diff --git a/.gitattributes b/.gitattributes
index 96eb6c4568ca5168f57c3f873b3353e2a3658889..02ecd02f5ff5aeec6a412fcebdcdd6e030b78424 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -69,3 +69,19 @@ L2/checkpoints/checkpoint-70/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 L2/checkpoints/checkpoint-80/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 L2/checkpoints/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 L3/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-10/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-100/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-110/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-120/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-130/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-140/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-150/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-159/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-20/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-30/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-40/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-50/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-60/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-70/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-80/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+L3/checkpoints/checkpoint-90/tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/L3/checkpoints/README.md b/L3/checkpoints/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..c9e1b45c3f2c163e22e44bb30165b4f69ede96ae
--- /dev/null
+++ b/L3/checkpoints/README.md
@@ -0,0 +1,58 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: transformers
+model_name: checkpoints
+tags:
+- generated_from_trainer
+- trl
+- sft
+licence: license
+---
+
+# Model Card for checkpoints
+
+This model is a fine-tuned version of [Qwen/Qwen3.5-0.8B](https://huggingface.co/Qwen/Qwen3.5-0.8B).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+ 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- TRL: 1.2.0.dev0
+- Transformers: 5.6.0.dev0
+- Pytorch: 2.9.0.dev20250803
+- Datasets: 4.8.4
+- Tokenizers: 0.22.1
+
+## Citations
+
+
+
+Cite TRL as:
+    
+```bibtex
+@software{vonwerra2020trl,
+  title   = {{TRL: Transformers Reinforcement Learning}},
+  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
+  license = {Apache-2.0},
+  url     = {https://github.com/huggingface/trl},
+  year    = {2020}
+}
+```
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-10/README.md b/L3/checkpoints/checkpoint-10/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-10/adapter_config.json b/L3/checkpoints/checkpoint-10/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-10/adapter_model.safetensors b/L3/checkpoints/checkpoint-10/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..81d67fc77eb464ea11c57e7c69df8bf3a81025b0
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:603831696ac5ebdf3453195533a72a9a729e1aa1c3d31a7051bba9f1b808298a
+size 4331744
diff --git a/L3/checkpoints/checkpoint-10/chat_template.jinja b/L3/checkpoints/checkpoint-10/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-10/optimizer.pt b/L3/checkpoints/checkpoint-10/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c901446567b32db58b224a55796bd01b387b20de
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70859cabf1091657539c71c239c675786cb06b43c92230c844f23f997815178a
+size 8690571
diff --git a/L3/checkpoints/checkpoint-10/rng_state.pth b/L3/checkpoints/checkpoint-10/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3f5c250a692e034a27fb06d7008612b2de078116
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cddf27219365242ec1046a3532a63a24c3f350c77f100e4f973369db2cc849d
+size 14455
diff --git a/L3/checkpoints/checkpoint-10/scheduler.pt b/L3/checkpoints/checkpoint-10/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f77a8c35a99260d65736ac6295c06ee1053ff084
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7fe1ca4b3f5360608595ce0b8c7a836143524c2bd8b15c3dedc0d66309506ab
+size 1465
diff --git a/L3/checkpoints/checkpoint-10/tokenizer.json b/L3/checkpoints/checkpoint-10/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-10/tokenizer_config.json b/L3/checkpoints/checkpoint-10/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-10/trainer_state.json b/L3/checkpoints/checkpoint-10/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..cf28e032401eb7a42fd6cdbb6befd4b0a722ff39
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/trainer_state.json
@@ -0,0 +1,44 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.18867924528301888,
+  "eval_steps": 500,
+  "global_step": 10,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 111779674536960.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-10/training_args.bin b/L3/checkpoints/checkpoint-10/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-10/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-100/README.md b/L3/checkpoints/checkpoint-100/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-100/adapter_config.json b/L3/checkpoints/checkpoint-100/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-100/adapter_model.safetensors b/L3/checkpoints/checkpoint-100/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1632be71ae8795ef08d86f0d09bf214171238669
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71b1ce6dd39084b482ad44bbd0b5951a309832203f1d5888a416d3698bbc1494
+size 4331744
diff --git a/L3/checkpoints/checkpoint-100/chat_template.jinja b/L3/checkpoints/checkpoint-100/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-100/optimizer.pt b/L3/checkpoints/checkpoint-100/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3c914db97c5592794f5b383e971d6c7d22c6a46e
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c97f4dee988dfa445d662fa06e172a84a15f21c1c541dd9b3d834b0e64ec04f
+size 8690571
diff --git a/L3/checkpoints/checkpoint-100/rng_state.pth b/L3/checkpoints/checkpoint-100/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..38c6d5df0a659b81300f358fefb31a6f4aa51ddc
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3a77d4a8b98ce027a4d6a3b9fb5d7c904e27ec1efd5c0468c24fa26bb738316
+size 14455
diff --git a/L3/checkpoints/checkpoint-100/scheduler.pt b/L3/checkpoints/checkpoint-100/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2e86f16121c94952328173d2bd55e9d7711b6ed2
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3177aafb877f9c541c793d0b5d70c393194524b75cdf776a7c9c47187919c5f0
+size 1465
diff --git a/L3/checkpoints/checkpoint-100/tokenizer.json b/L3/checkpoints/checkpoint-100/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-100/tokenizer_config.json b/L3/checkpoints/checkpoint-100/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-100/trainer_state.json b/L3/checkpoints/checkpoint-100/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..83a145a48f295adb9653d3fb2d5dc8ffc6487676
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/trainer_state.json
@@ -0,0 +1,134 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.8867924528301887,
+  "eval_steps": 500,
+  "global_step": 100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.33994140625,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7750918865203857,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.3485977172851562,
+      "mean_token_accuracy": 0.7031819194555282,
+      "num_tokens": 361755.0,
+      "step": 100
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1133377609971456.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-100/training_args.bin b/L3/checkpoints/checkpoint-100/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-100/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-110/README.md b/L3/checkpoints/checkpoint-110/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-110/adapter_config.json b/L3/checkpoints/checkpoint-110/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-110/adapter_model.safetensors b/L3/checkpoints/checkpoint-110/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7a2ec31fd65309d37a7e99687fa13a0a80f89cc4
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a05fabafa5405733ebe5187361867cf15a23dd838bb3ea25a262aede051fdaf
+size 4331744
diff --git a/L3/checkpoints/checkpoint-110/chat_template.jinja b/L3/checkpoints/checkpoint-110/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-110/optimizer.pt b/L3/checkpoints/checkpoint-110/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..135216e45becc4b7b50d073fa5764dc9cead6d11
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36a133b64929e209adce564e1fa334cf33184b8e921b738fa283d0fa7ac792f2
+size 8690571
diff --git a/L3/checkpoints/checkpoint-110/rng_state.pth b/L3/checkpoints/checkpoint-110/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7d150fc4504460cdea273ef2adcb04a65c626709
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba4fde4ee04d2f472bb4dea96a48e8fdf7891d2b0694a8f012e8133a2e176ae
+size 14455
diff --git a/L3/checkpoints/checkpoint-110/scheduler.pt b/L3/checkpoints/checkpoint-110/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2a9fadb38e19952787f70f5d92bdb6eeb301a08a
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8f882f0bf7e74a2b304dde78b9fa019bc2cdbf9058f544ba3b9a9ae35031b93
+size 1465
diff --git a/L3/checkpoints/checkpoint-110/tokenizer.json b/L3/checkpoints/checkpoint-110/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-110/tokenizer_config.json b/L3/checkpoints/checkpoint-110/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-110/trainer_state.json b/L3/checkpoints/checkpoint-110/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..dd68303bba3d7607eb91094f128aecf452324f6e
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/trainer_state.json
@@ -0,0 +1,144 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0754716981132075,
+  "eval_steps": 500,
+  "global_step": 110,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.33994140625,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7750918865203857,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.3485977172851562,
+      "mean_token_accuracy": 0.7031819194555282,
+      "num_tokens": 361755.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.32060546875,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.6582154631614685,
+      "learning_rate": 4.8811495091039926e-05,
+      "loss": 1.3079830169677735,
+      "mean_token_accuracy": 0.7036498673260212,
+      "num_tokens": 398143.0,
+      "step": 110
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1246594965209856.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-110/training_args.bin b/L3/checkpoints/checkpoint-110/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-110/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-120/README.md b/L3/checkpoints/checkpoint-120/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-120/adapter_config.json b/L3/checkpoints/checkpoint-120/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-120/adapter_model.safetensors b/L3/checkpoints/checkpoint-120/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..41d3474f45f263981c2d4b610de4b5a867a7fafd
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37d8bb647cd39430b378d4f51e47e77e8914814cf930728cb46f9a54ce44abaf
+size 4331744
diff --git a/L3/checkpoints/checkpoint-120/chat_template.jinja b/L3/checkpoints/checkpoint-120/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-120/optimizer.pt b/L3/checkpoints/checkpoint-120/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fa0437e1f22e7e63758a44e85904247eac36b8f5
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c30420b12e6feb1067d57c398bcc4085dc68457f875ece9fd9254c5a7b33ff64
+size 8690571
diff --git a/L3/checkpoints/checkpoint-120/rng_state.pth b/L3/checkpoints/checkpoint-120/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7d150fc4504460cdea273ef2adcb04a65c626709
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba4fde4ee04d2f472bb4dea96a48e8fdf7891d2b0694a8f012e8133a2e176ae
+size 14455
diff --git a/L3/checkpoints/checkpoint-120/scheduler.pt b/L3/checkpoints/checkpoint-120/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9360bd36ac51b5e1d4ce1f2a58618eee24bcb300
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c748cde43f3ac780a033e33735b44512b14c6bcfd555ce64d6b760d4ce31be08
+size 1465
diff --git a/L3/checkpoints/checkpoint-120/tokenizer.json b/L3/checkpoints/checkpoint-120/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-120/tokenizer_config.json b/L3/checkpoints/checkpoint-120/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-120/trainer_state.json b/L3/checkpoints/checkpoint-120/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c9df4c51d3c799609d347f24caf404b321b7ca37
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/trainer_state.json
@@ -0,0 +1,154 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.2641509433962264,
+  "eval_steps": 500,
+  "global_step": 120,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.33994140625,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7750918865203857,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.3485977172851562,
+      "mean_token_accuracy": 0.7031819194555282,
+      "num_tokens": 361755.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.32060546875,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.6582154631614685,
+      "learning_rate": 4.8811495091039926e-05,
+      "loss": 1.3079830169677735,
+      "mean_token_accuracy": 0.7036498673260212,
+      "num_tokens": 398143.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.31796875,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.6963320374488831,
+      "learning_rate": 3.2271842837425906e-05,
+      "loss": 1.312546443939209,
+      "mean_token_accuracy": 0.704971169680357,
+      "num_tokens": 434299.0,
+      "step": 120
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1359902175492096.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-120/training_args.bin b/L3/checkpoints/checkpoint-120/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-120/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-130/README.md b/L3/checkpoints/checkpoint-130/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-130/adapter_config.json b/L3/checkpoints/checkpoint-130/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-130/adapter_model.safetensors b/L3/checkpoints/checkpoint-130/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..204cfb79988055eee1ad89c30c3e392283c865cd
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4646c8698a341f8e4c5f92512b14d51a01699351beab6b63e315e8ba79506403
+size 4331744
diff --git a/L3/checkpoints/checkpoint-130/chat_template.jinja b/L3/checkpoints/checkpoint-130/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-130/optimizer.pt b/L3/checkpoints/checkpoint-130/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..66a6bf70c75b086ea6858eb79dfa0feeb7578ceb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a51c2129c35709656801ad7286adf84aed6dcda5558b6c2abff73c77375dcefe
+size 8690571
diff --git a/L3/checkpoints/checkpoint-130/rng_state.pth b/L3/checkpoints/checkpoint-130/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7d150fc4504460cdea273ef2adcb04a65c626709
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba4fde4ee04d2f472bb4dea96a48e8fdf7891d2b0694a8f012e8133a2e176ae
+size 14455
diff --git a/L3/checkpoints/checkpoint-130/scheduler.pt b/L3/checkpoints/checkpoint-130/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5b6b9c835319c32c5ba395960f5913e2075263be
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e84db9e363097911357934d81595bf7aeeaa4a6e8b69e7eeb5390b4b7770cc56
+size 1465
diff --git a/L3/checkpoints/checkpoint-130/tokenizer.json b/L3/checkpoints/checkpoint-130/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-130/tokenizer_config.json b/L3/checkpoints/checkpoint-130/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-130/trainer_state.json b/L3/checkpoints/checkpoint-130/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..dbec252f3fae05079cdb7cfa9d030148b4d1aee1
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/trainer_state.json
@@ -0,0 +1,164 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.452830188679245,
+  "eval_steps": 500,
+  "global_step": 130,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.33994140625,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7750918865203857,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.3485977172851562,
+      "mean_token_accuracy": 0.7031819194555282,
+      "num_tokens": 361755.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.32060546875,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.6582154631614685,
+      "learning_rate": 4.8811495091039926e-05,
+      "loss": 1.3079830169677735,
+      "mean_token_accuracy": 0.7036498673260212,
+      "num_tokens": 398143.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.31796875,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.6963320374488831,
+      "learning_rate": 3.2271842837425906e-05,
+      "loss": 1.312546443939209,
+      "mean_token_accuracy": 0.704971169680357,
+      "num_tokens": 434299.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.32119140625,
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.6907761096954346,
+      "learning_rate": 1.861512827298051e-05,
+      "loss": 1.3149855613708497,
+      "mean_token_accuracy": 0.7058135308325291,
+      "num_tokens": 470708.0,
+      "step": 130
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1474533250086912.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-130/training_args.bin b/L3/checkpoints/checkpoint-130/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-130/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-140/README.md b/L3/checkpoints/checkpoint-140/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-140/adapter_config.json b/L3/checkpoints/checkpoint-140/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-140/adapter_model.safetensors b/L3/checkpoints/checkpoint-140/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f76dd85e6d044b4dc72461a3e0d07f4f7509f332
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac7e7aebc1bd00f196f490a535233b501a5125c330627189206d3f983ac2604c
+size 4331744
diff --git a/L3/checkpoints/checkpoint-140/chat_template.jinja b/L3/checkpoints/checkpoint-140/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-140/optimizer.pt b/L3/checkpoints/checkpoint-140/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..61855a5071a7c79c077a9a0e3097cb385d407849
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7c26169c7c3a3ecd2a3e8811ca3d03513cd5d4aac45da38f51c86f8c4572a76
+size 8690571
diff --git a/L3/checkpoints/checkpoint-140/rng_state.pth b/L3/checkpoints/checkpoint-140/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7d150fc4504460cdea273ef2adcb04a65c626709
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba4fde4ee04d2f472bb4dea96a48e8fdf7891d2b0694a8f012e8133a2e176ae
+size 14455
diff --git a/L3/checkpoints/checkpoint-140/scheduler.pt b/L3/checkpoints/checkpoint-140/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b49e885f7ca1f34c0094cc8f287cae0d005282f0
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfa9e2ffd448b5ae0809078fa803005d3fb0168890bd12b132bdb4cb8a1d4596
+size 1465
diff --git a/L3/checkpoints/checkpoint-140/tokenizer.json b/L3/checkpoints/checkpoint-140/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-140/tokenizer_config.json b/L3/checkpoints/checkpoint-140/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-140/trainer_state.json b/L3/checkpoints/checkpoint-140/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ad4a127143395244579dda0da268c595491b4a6c
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/trainer_state.json
@@ -0,0 +1,174 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.641509433962264,
+  "eval_steps": 500,
+  "global_step": 140,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.33994140625,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7750918865203857,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.3485977172851562,
+      "mean_token_accuracy": 0.7031819194555282,
+      "num_tokens": 361755.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.32060546875,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.6582154631614685,
+      "learning_rate": 4.8811495091039926e-05,
+      "loss": 1.3079830169677735,
+      "mean_token_accuracy": 0.7036498673260212,
+      "num_tokens": 398143.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.31796875,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.6963320374488831,
+      "learning_rate": 3.2271842837425906e-05,
+      "loss": 1.312546443939209,
+      "mean_token_accuracy": 0.704971169680357,
+      "num_tokens": 434299.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.32119140625,
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.6907761096954346,
+      "learning_rate": 1.861512827298051e-05,
+      "loss": 1.3149855613708497,
+      "mean_token_accuracy": 0.7058135308325291,
+      "num_tokens": 470708.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.33837890625,
+      "epoch": 2.641509433962264,
+      "grad_norm": 0.7412691712379456,
+      "learning_rate": 8.422667334494249e-06,
+      "loss": 1.3250213623046876,
+      "mean_token_accuracy": 0.7025255858898163,
+      "num_tokens": 506485.0,
+      "step": 140
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1586055340164864.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-140/training_args.bin b/L3/checkpoints/checkpoint-140/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-140/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-150/README.md b/L3/checkpoints/checkpoint-150/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-150/adapter_config.json b/L3/checkpoints/checkpoint-150/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-150/adapter_model.safetensors b/L3/checkpoints/checkpoint-150/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..93d0c1225d0adb7e77d1acae2c523e6e75cc6bc7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19f1f6caaf3b2df1e40b12d9af9f8c8fd649f3472813c6dc7c149e9a9b56011c
+size 4331744
diff --git a/L3/checkpoints/checkpoint-150/chat_template.jinja b/L3/checkpoints/checkpoint-150/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-150/optimizer.pt b/L3/checkpoints/checkpoint-150/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..95b9e1682cd1b8f9629f63f2e9669f0307813c53
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af621c8dd054bd157ea26a9feb961e6a7681285a3731a0ff8d8dc4547591b66a
+size 8690571
diff --git a/L3/checkpoints/checkpoint-150/rng_state.pth b/L3/checkpoints/checkpoint-150/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7d150fc4504460cdea273ef2adcb04a65c626709
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba4fde4ee04d2f472bb4dea96a48e8fdf7891d2b0694a8f012e8133a2e176ae
+size 14455
diff --git a/L3/checkpoints/checkpoint-150/scheduler.pt b/L3/checkpoints/checkpoint-150/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..08f0521496768738e3855a6425ce4776abf8262b
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:749a70a739231624cd72b7c431bbcfa6d3f5dea58efe86a276df4a9d9b201e64
+size 1465
diff --git a/L3/checkpoints/checkpoint-150/tokenizer.json b/L3/checkpoints/checkpoint-150/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-150/tokenizer_config.json b/L3/checkpoints/checkpoint-150/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-150/trainer_state.json b/L3/checkpoints/checkpoint-150/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..10a9cd44493f80e80c1e8b1454987836f3a32cb7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/trainer_state.json
@@ -0,0 +1,184 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.830188679245283,
+  "eval_steps": 500,
+  "global_step": 150,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.33994140625,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7750918865203857,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.3485977172851562,
+      "mean_token_accuracy": 0.7031819194555282,
+      "num_tokens": 361755.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.32060546875,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.6582154631614685,
+      "learning_rate": 4.8811495091039926e-05,
+      "loss": 1.3079830169677735,
+      "mean_token_accuracy": 0.7036498673260212,
+      "num_tokens": 398143.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.31796875,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.6963320374488831,
+      "learning_rate": 3.2271842837425906e-05,
+      "loss": 1.312546443939209,
+      "mean_token_accuracy": 0.704971169680357,
+      "num_tokens": 434299.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.32119140625,
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.6907761096954346,
+      "learning_rate": 1.861512827298051e-05,
+      "loss": 1.3149855613708497,
+      "mean_token_accuracy": 0.7058135308325291,
+      "num_tokens": 470708.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.33837890625,
+      "epoch": 2.641509433962264,
+      "grad_norm": 0.7412691712379456,
+      "learning_rate": 8.422667334494249e-06,
+      "loss": 1.3250213623046876,
+      "mean_token_accuracy": 0.7025255858898163,
+      "num_tokens": 506485.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.33466796875,
+      "epoch": 2.830188679245283,
+      "grad_norm": 0.7045652866363525,
+      "learning_rate": 2.128315467264552e-06,
+      "loss": 1.324216938018799,
+      "mean_token_accuracy": 0.7067199409008026,
+      "num_tokens": 542651.0,
+      "step": 150
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1698943226909184.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-150/training_args.bin b/L3/checkpoints/checkpoint-150/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-150/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-159/README.md b/L3/checkpoints/checkpoint-159/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-159/adapter_config.json b/L3/checkpoints/checkpoint-159/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-159/adapter_model.safetensors b/L3/checkpoints/checkpoint-159/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..9b52576537b3da06122991888102681ed89c09a5
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98a33b83c8b5cc332a5005653cc47eb92f1ff2e9f1be2bd56a8d2f8df58ba93a
+size 4331744
diff --git a/L3/checkpoints/checkpoint-159/chat_template.jinja b/L3/checkpoints/checkpoint-159/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-159/optimizer.pt b/L3/checkpoints/checkpoint-159/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3eae43175d191cd7bb1800044a729ffe6ed95a78
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d26df2b135af6602a11c180ce0c6fffd9338d05116274c7db5aff248b16aed6
+size 8690571
diff --git a/L3/checkpoints/checkpoint-159/rng_state.pth b/L3/checkpoints/checkpoint-159/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7d150fc4504460cdea273ef2adcb04a65c626709
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba4fde4ee04d2f472bb4dea96a48e8fdf7891d2b0694a8f012e8133a2e176ae
+size 14455
diff --git a/L3/checkpoints/checkpoint-159/scheduler.pt b/L3/checkpoints/checkpoint-159/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f87e3c19ec47a9a829bf65c0292601f71a3ae173
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b2c6206a97d0884fd8dcbde955ce8b12473b8e47fac73a0c0671e5d2c6dfdd9
+size 1465
diff --git a/L3/checkpoints/checkpoint-159/tokenizer.json b/L3/checkpoints/checkpoint-159/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-159/tokenizer_config.json b/L3/checkpoints/checkpoint-159/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-159/trainer_state.json b/L3/checkpoints/checkpoint-159/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..cb0eb97520e4a444ebf9709ffd2f9ac5b5908cf4
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/trainer_state.json
@@ -0,0 +1,184 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 159,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.33994140625,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7750918865203857,
+      "learning_rate": 6.753005307953167e-05,
+      "loss": 1.3485977172851562,
+      "mean_token_accuracy": 0.7031819194555282,
+      "num_tokens": 361755.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.32060546875,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.6582154631614685,
+      "learning_rate": 4.8811495091039926e-05,
+      "loss": 1.3079830169677735,
+      "mean_token_accuracy": 0.7036498673260212,
+      "num_tokens": 398143.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.31796875,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.6963320374488831,
+      "learning_rate": 3.2271842837425906e-05,
+      "loss": 1.312546443939209,
+      "mean_token_accuracy": 0.704971169680357,
+      "num_tokens": 434299.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.32119140625,
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.6907761096954346,
+      "learning_rate": 1.861512827298051e-05,
+      "loss": 1.3149855613708497,
+      "mean_token_accuracy": 0.7058135308325291,
+      "num_tokens": 470708.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.33837890625,
+      "epoch": 2.641509433962264,
+      "grad_norm": 0.7412691712379456,
+      "learning_rate": 8.422667334494249e-06,
+      "loss": 1.3250213623046876,
+      "mean_token_accuracy": 0.7025255858898163,
+      "num_tokens": 506485.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.33466796875,
+      "epoch": 2.830188679245283,
+      "grad_norm": 0.7045652866363525,
+      "learning_rate": 2.128315467264552e-06,
+      "loss": 1.324216938018799,
+      "mean_token_accuracy": 0.7067199409008026,
+      "num_tokens": 542651.0,
+      "step": 150
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1801377976886784.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-159/training_args.bin b/L3/checkpoints/checkpoint-159/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-159/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-20/README.md b/L3/checkpoints/checkpoint-20/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-20/adapter_config.json b/L3/checkpoints/checkpoint-20/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-20/adapter_model.safetensors b/L3/checkpoints/checkpoint-20/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ab0f1f55b0b29632babf236419b704abf97836a1
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c92ebacd0594bcb79afd1849c3f33852165a6706aac4affbfe71566d22a8247e
+size 4331744
diff --git a/L3/checkpoints/checkpoint-20/chat_template.jinja b/L3/checkpoints/checkpoint-20/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-20/optimizer.pt b/L3/checkpoints/checkpoint-20/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..805482641332574670ae1720197013d8b6a103c5
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:394bbcf3988133d52c965de3bc60c7232e8bc46f25221ef32d32eb05cd03e867
+size 8690571
diff --git a/L3/checkpoints/checkpoint-20/rng_state.pth b/L3/checkpoints/checkpoint-20/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3f5c250a692e034a27fb06d7008612b2de078116
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cddf27219365242ec1046a3532a63a24c3f350c77f100e4f973369db2cc849d
+size 14455
diff --git a/L3/checkpoints/checkpoint-20/scheduler.pt b/L3/checkpoints/checkpoint-20/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..19c5d720d9fd1aadfddfa1592620d1a5c18e2d01
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a55c2265171097c74232991dbd82262b6cfcfa2cf3bd4b0d5a7dfec293bf8656
+size 1465
diff --git a/L3/checkpoints/checkpoint-20/tokenizer.json b/L3/checkpoints/checkpoint-20/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-20/tokenizer_config.json b/L3/checkpoints/checkpoint-20/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-20/trainer_state.json b/L3/checkpoints/checkpoint-20/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..7a7664f463a24117f2b514b44a4849b3baad64be
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/trainer_state.json
@@ -0,0 +1,54 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.37735849056603776,
+  "eval_steps": 500,
+  "global_step": 20,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 224931136076544.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-20/training_args.bin b/L3/checkpoints/checkpoint-20/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-20/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-30/README.md b/L3/checkpoints/checkpoint-30/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-30/adapter_config.json b/L3/checkpoints/checkpoint-30/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-30/adapter_model.safetensors b/L3/checkpoints/checkpoint-30/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..58e2f049fc6658f24bd2954502fbe82f0060e2ff
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c066166f8f49ec92950334c86c328114521ea628b6966d391512cb5a72d0f304
+size 4331744
diff --git a/L3/checkpoints/checkpoint-30/chat_template.jinja b/L3/checkpoints/checkpoint-30/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-30/optimizer.pt b/L3/checkpoints/checkpoint-30/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..541b793a1599ac45908e6ea3b3595bf25588afd0
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9369b0eecd030002437e9c4a6b7bbee49455884aa60d35b5979c377efd375a7f
+size 8690571
diff --git a/L3/checkpoints/checkpoint-30/rng_state.pth b/L3/checkpoints/checkpoint-30/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3f5c250a692e034a27fb06d7008612b2de078116
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cddf27219365242ec1046a3532a63a24c3f350c77f100e4f973369db2cc849d
+size 14455
diff --git a/L3/checkpoints/checkpoint-30/scheduler.pt b/L3/checkpoints/checkpoint-30/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e67f345290edd2767d65bc2a0d39f112608d731e
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5714db383b01493e282f902afe194d44c50334403abd999cb0a6f6d875dd5b8
+size 1465
diff --git a/L3/checkpoints/checkpoint-30/tokenizer.json b/L3/checkpoints/checkpoint-30/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-30/tokenizer_config.json b/L3/checkpoints/checkpoint-30/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-30/trainer_state.json b/L3/checkpoints/checkpoint-30/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e1e8663b9bc03d5b121b6eac469a6712e5d542a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/trainer_state.json
@@ -0,0 +1,64 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5660377358490566,
+  "eval_steps": 500,
+  "global_step": 30,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 338388104765184.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-30/training_args.bin b/L3/checkpoints/checkpoint-30/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-30/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-40/README.md b/L3/checkpoints/checkpoint-40/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-40/adapter_config.json b/L3/checkpoints/checkpoint-40/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-40/adapter_model.safetensors b/L3/checkpoints/checkpoint-40/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4fa59fb84e2fe706210d0583c632b1195b28cdaa
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f52694653d39f675397a2977fdf947b489e9846bd043a68008363f271581b193
+size 4331744
diff --git a/L3/checkpoints/checkpoint-40/chat_template.jinja b/L3/checkpoints/checkpoint-40/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-40/optimizer.pt b/L3/checkpoints/checkpoint-40/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9a877a3ede0cbbdc933a9c096b83d810d816bfbf
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:546642515f51f166a34576368a62c0cda729fe54f83adeb207c834c5b1ec9d9e
+size 8690571
diff --git a/L3/checkpoints/checkpoint-40/rng_state.pth b/L3/checkpoints/checkpoint-40/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3f5c250a692e034a27fb06d7008612b2de078116
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cddf27219365242ec1046a3532a63a24c3f350c77f100e4f973369db2cc849d
+size 14455
diff --git a/L3/checkpoints/checkpoint-40/scheduler.pt b/L3/checkpoints/checkpoint-40/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d277614299656d3c5f7cf2fa488957961f886c69
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6ab065d4e48130a57f256525f3b5c7ec35cb4454ae9ef2f4ce13133dbee5eee
+size 1465
diff --git a/L3/checkpoints/checkpoint-40/tokenizer.json b/L3/checkpoints/checkpoint-40/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-40/tokenizer_config.json b/L3/checkpoints/checkpoint-40/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-40/trainer_state.json b/L3/checkpoints/checkpoint-40/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..79517bc1e3520207fe3474937825bed2dedb314d
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/trainer_state.json
@@ -0,0 +1,74 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7547169811320755,
+  "eval_steps": 500,
+  "global_step": 40,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 452432126406912.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-40/training_args.bin b/L3/checkpoints/checkpoint-40/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-40/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-50/README.md b/L3/checkpoints/checkpoint-50/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-50/adapter_config.json b/L3/checkpoints/checkpoint-50/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-50/adapter_model.safetensors b/L3/checkpoints/checkpoint-50/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cfc737608add87168ca02486a8293fb60763e8c2
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b20e81fb6669c223c2ccc6b6a71f0c11ec7bddf7eedca19bb673f18cc36457c4
+size 4331744
diff --git a/L3/checkpoints/checkpoint-50/chat_template.jinja b/L3/checkpoints/checkpoint-50/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-50/optimizer.pt b/L3/checkpoints/checkpoint-50/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2c3b5bff51e1b619d17277ca289b857172843e6d
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:037263d94c731c80f48006b55cc09d727765c7e7f4c27e2285e32dbb763de161
+size 8690571
diff --git a/L3/checkpoints/checkpoint-50/rng_state.pth b/L3/checkpoints/checkpoint-50/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3f5c250a692e034a27fb06d7008612b2de078116
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cddf27219365242ec1046a3532a63a24c3f350c77f100e4f973369db2cc849d
+size 14455
diff --git a/L3/checkpoints/checkpoint-50/scheduler.pt b/L3/checkpoints/checkpoint-50/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8dfa70f8756ce63cbbdda959a0f49cb5d5040eaf
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06e42634db8916928265bbd702cb453e8ac4774a2c83e1b2f1905185363111e5
+size 1465
diff --git a/L3/checkpoints/checkpoint-50/tokenizer.json b/L3/checkpoints/checkpoint-50/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-50/tokenizer_config.json b/L3/checkpoints/checkpoint-50/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-50/trainer_state.json b/L3/checkpoints/checkpoint-50/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..5cd232bf8a5d3ff282ce23bb34b1814edce28a33
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/trainer_state.json
@@ -0,0 +1,84 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9433962264150944,
+  "eval_steps": 500,
+  "global_step": 50,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 566422235022336.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-50/training_args.bin b/L3/checkpoints/checkpoint-50/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-50/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-60/README.md b/L3/checkpoints/checkpoint-60/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-60/adapter_config.json b/L3/checkpoints/checkpoint-60/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-60/adapter_model.safetensors b/L3/checkpoints/checkpoint-60/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6c3c8e6bbe0a8f6e600ea492e85af4422d33fe16
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f645f95f1f8ec61dee2d2e39680b98934a9308a0ba90bc4ad46220287e9c2ca6
+size 4331744
diff --git a/L3/checkpoints/checkpoint-60/chat_template.jinja b/L3/checkpoints/checkpoint-60/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-60/optimizer.pt b/L3/checkpoints/checkpoint-60/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ca0ac9bad3385f578364efad992a71c209f9188f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f36549f13122b556101336cd43afb03c7f31bc5a49385a47818d45b4e14dffc0
+size 8690571
diff --git a/L3/checkpoints/checkpoint-60/rng_state.pth b/L3/checkpoints/checkpoint-60/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..38c6d5df0a659b81300f358fefb31a6f4aa51ddc
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3a77d4a8b98ce027a4d6a3b9fb5d7c904e27ec1efd5c0468c24fa26bb738316
+size 14455
diff --git a/L3/checkpoints/checkpoint-60/scheduler.pt b/L3/checkpoints/checkpoint-60/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0332598581c0fc180611ee71c3947361daad8502
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78d647f03c1cf87f7118cf942735589fe5fc5b2f1991d453c311ad9bb47c1543
+size 1465
diff --git a/L3/checkpoints/checkpoint-60/tokenizer.json b/L3/checkpoints/checkpoint-60/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-60/tokenizer_config.json b/L3/checkpoints/checkpoint-60/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-60/trainer_state.json b/L3/checkpoints/checkpoint-60/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..7e26c14ad52a296e601047a624c2a1c9cb702ead
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/trainer_state.json
@@ -0,0 +1,94 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.1320754716981132,
+  "eval_steps": 500,
+  "global_step": 60,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 679483841518080.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-60/training_args.bin b/L3/checkpoints/checkpoint-60/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-60/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-70/README.md b/L3/checkpoints/checkpoint-70/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-70/adapter_config.json b/L3/checkpoints/checkpoint-70/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-70/adapter_model.safetensors b/L3/checkpoints/checkpoint-70/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..eca3041daa0510f1ad1bb43727c63615dcefc92e
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09f01de7f2e0dd75703483d62784de6372d48c1fa801d66cf1d33c8201f08bd6
+size 4331744
diff --git a/L3/checkpoints/checkpoint-70/chat_template.jinja b/L3/checkpoints/checkpoint-70/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-70/optimizer.pt b/L3/checkpoints/checkpoint-70/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..848bd2e6e8acc0c780afb1c179999be304461f2c
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b19263c77724f4540eda65c53e851fa34449df1964288e2a67d43253d07da20a
+size 8690571
diff --git a/L3/checkpoints/checkpoint-70/rng_state.pth b/L3/checkpoints/checkpoint-70/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..38c6d5df0a659b81300f358fefb31a6f4aa51ddc
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3a77d4a8b98ce027a4d6a3b9fb5d7c904e27ec1efd5c0468c24fa26bb738316
+size 14455
diff --git a/L3/checkpoints/checkpoint-70/scheduler.pt b/L3/checkpoints/checkpoint-70/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f858bbf27b38abd1e565ff130e1a53dec875216b
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a6fd9f3206f85fc00865417aaeb161c58ccaf5e621cc80ea7f1ca566f53f390
+size 1465
diff --git a/L3/checkpoints/checkpoint-70/tokenizer.json b/L3/checkpoints/checkpoint-70/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-70/tokenizer_config.json b/L3/checkpoints/checkpoint-70/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-70/trainer_state.json b/L3/checkpoints/checkpoint-70/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..39cd551eddf15eafa9fb9d02e776807d5a667732
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/trainer_state.json
@@ -0,0 +1,104 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.320754716981132,
+  "eval_steps": 500,
+  "global_step": 70,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 793599747194880.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-70/training_args.bin b/L3/checkpoints/checkpoint-70/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-70/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-80/README.md b/L3/checkpoints/checkpoint-80/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-80/adapter_config.json b/L3/checkpoints/checkpoint-80/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-80/adapter_model.safetensors b/L3/checkpoints/checkpoint-80/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a3bdb0b7ef8a7e9bf03852518776b17b0a3e3da4
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c44ca2879de762b27a982037bdf90ba46201109da8ac59d22893ad208db0f221
+size 4331744
diff --git a/L3/checkpoints/checkpoint-80/chat_template.jinja b/L3/checkpoints/checkpoint-80/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-80/optimizer.pt b/L3/checkpoints/checkpoint-80/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3388974ef0c35316caa5270b207e1c4ef520d632
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66eb8dd952d9612f9ddaa431576898db5d699a5eaa847a204d31be31267ad6a0
+size 8690571
diff --git a/L3/checkpoints/checkpoint-80/rng_state.pth b/L3/checkpoints/checkpoint-80/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..38c6d5df0a659b81300f358fefb31a6f4aa51ddc
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3a77d4a8b98ce027a4d6a3b9fb5d7c904e27ec1efd5c0468c24fa26bb738316
+size 14455
diff --git a/L3/checkpoints/checkpoint-80/scheduler.pt b/L3/checkpoints/checkpoint-80/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0b246dec42e48802d2bbbafbff44c90ed32e5ce1
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a56e71ec595f148f31173cc2e0fe75a40a9eb1e044fd9c879295d92c4f269c83
+size 1465
diff --git a/L3/checkpoints/checkpoint-80/tokenizer.json b/L3/checkpoints/checkpoint-80/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-80/tokenizer_config.json b/L3/checkpoints/checkpoint-80/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-80/trainer_state.json b/L3/checkpoints/checkpoint-80/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..2db1f14f94344d352f07c2e2b1b0c32dd54cec29
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/trainer_state.json
@@ -0,0 +1,114 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.509433962264151,
+  "eval_steps": 500,
+  "global_step": 80,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 905301547360512.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-80/training_args.bin b/L3/checkpoints/checkpoint-80/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-80/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713
diff --git a/L3/checkpoints/checkpoint-90/README.md b/L3/checkpoints/checkpoint-90/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..9b1d0d5c752a5c9f2e3a0e3a9ead3759650a12a9
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-0.8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-0.8B
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.2.dev0
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-90/adapter_config.json b/L3/checkpoints/checkpoint-90/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..466b55ef5352db25f2ac04e8aec438ce0764d848
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/adapter_config.json
@@ -0,0 +1,45 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-0.8B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.2.dev0@076214c61f690898509b97702b5e9d95c826f000",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-90/adapter_model.safetensors b/L3/checkpoints/checkpoint-90/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..30bc05165b0d72dd9b43853f6abcf290dd802494
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e15af3a94f1a6f99970e419b8616418d28b7103e91c3669f40067cc9632f5ce
+size 4331744
diff --git a/L3/checkpoints/checkpoint-90/chat_template.jinja b/L3/checkpoints/checkpoint-90/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..0ef09f214eaa6d9bca297988afc1454b5827b2c7
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/chat_template.jinja
@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/L3/checkpoints/checkpoint-90/optimizer.pt b/L3/checkpoints/checkpoint-90/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b3dac5f08adaa28512f6d40e832653a96c48ca37
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b4505fbd04ed43a2014f967399c260400e4945bdc2d04e1750f30d16b630f06
+size 8690571
diff --git a/L3/checkpoints/checkpoint-90/rng_state.pth b/L3/checkpoints/checkpoint-90/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..38c6d5df0a659b81300f358fefb31a6f4aa51ddc
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3a77d4a8b98ce027a4d6a3b9fb5d7c904e27ec1efd5c0468c24fa26bb738316
+size 14455
diff --git a/L3/checkpoints/checkpoint-90/scheduler.pt b/L3/checkpoints/checkpoint-90/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9996ef92f3da1dcc1df8c7e09ce0aac1f761217a
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:111d012cb61279f251d1c7420ef60a994ecb49ac992cc1aad87f183316311861
+size 1465
diff --git a/L3/checkpoints/checkpoint-90/tokenizer.json b/L3/checkpoints/checkpoint-90/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..67741b04f23bfdb46501f748ce27865ec82eccfb
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4
+size 19989343
diff --git a/L3/checkpoints/checkpoint-90/tokenizer_config.json b/L3/checkpoints/checkpoint-90/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0871602fcb7d15fcb2979685b1c7b609eef01a6f
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/tokenizer_config.json
@@ -0,0 +1,32 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/L3/checkpoints/checkpoint-90/trainer_state.json b/L3/checkpoints/checkpoint-90/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c929a982d3436cf55a6709909b54cc9c39b9f74b
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/trainer_state.json
@@ -0,0 +1,124 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.6981132075471699,
+  "eval_steps": 500,
+  "global_step": 90,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.27734375,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.4095726013183594,
+      "learning_rate": 0.0001999145758387301,
+      "loss": 2.535823440551758,
+      "mean_token_accuracy": 0.4902416098862886,
+      "num_tokens": 36112.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.03623046875,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 1.8039652109146118,
+      "learning_rate": 0.00019694002659393305,
+      "loss": 1.8651376724243165,
+      "mean_token_accuracy": 0.5952314671128989,
+      "num_tokens": 72245.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.49443359375,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7885380387306213,
+      "learning_rate": 0.0001898390981891979,
+      "loss": 1.5259657859802247,
+      "mean_token_accuracy": 0.6693046040832996,
+      "num_tokens": 108540.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.421875,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.7279834747314453,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 1.4288103103637695,
+      "mean_token_accuracy": 0.6861812733113766,
+      "num_tokens": 144637.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.38642578125,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.6967478394508362,
+      "learning_rate": 0.00016462992378609407,
+      "loss": 1.3759333610534668,
+      "mean_token_accuracy": 0.6965034589171409,
+      "num_tokens": 180890.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.37880859375,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.7058449983596802,
+      "learning_rate": 0.00014759473930370736,
+      "loss": 1.3774614334106445,
+      "mean_token_accuracy": 0.6968415275216102,
+      "num_tokens": 216977.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.3544921875,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7576785087585449,
+      "learning_rate": 0.00012853362242491053,
+      "loss": 1.3501564979553222,
+      "mean_token_accuracy": 0.699217775464058,
+      "num_tokens": 253285.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.35498046875,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.6370306015014648,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 1.346321201324463,
+      "mean_token_accuracy": 0.7007117375731469,
+      "num_tokens": 288932.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.362890625,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7341946363449097,
+      "learning_rate": 8.763073687306524e-05,
+      "loss": 1.3396063804626466,
+      "mean_token_accuracy": 0.701534990221262,
+      "num_tokens": 325309.0,
+      "step": 90
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 159,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1018884313110528.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/L3/checkpoints/checkpoint-90/training_args.bin b/L3/checkpoints/checkpoint-90/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d669276de59e9e4d7c51433f618d2e491cd91256
--- /dev/null
+++ b/L3/checkpoints/checkpoint-90/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80b9bb2dae65508f4e6141fa8a45b8dc53f9794f9c5757bce2273b5609882b14
+size 5713