Add files using upload-large-folder tool

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+model.safetensors.index.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,225 @@
+---
+library_name: transformers
+license: mit
+pipeline_tag: text-generation
+tags:
+- vLLM
+- AWQ
+base_model:
+  - zai-org/GLM-5
+base_model_relation: quantized
+
+---
+# GLM-5-AWQ
+Base model: [zai-org/GLM-5](https://huggingface.co/zai-org/GLM-5)
+
+This repo quantizes the model using data-free quantization (no calibration dataset required).
+
+### 【Dependencies / Installation】
+
+```python
+# NOTE: 
+# vllm==0.16.0rc2 absolutely would NOT work! 
+# Must upgrade to >=0.16.1rc1
+vllm>=0.16.1rc1.dev7    
+transformers>=5.3.0.dev0
+```
+
+As of **2026-02-26**, make sure your system has cuda12.8 installed.
+
+Then, create a fresh Python environment (e.g. python3.12 venv) and run:
+```bash
+pip install -U vllm --pre --index-url https://pypi.org/simple --extra-index-url https://wheels.vllm.ai/nightly
+pip install git+https://github.com/huggingface/transformers.git
+pip install git+https://github.com/deepseek-ai/DeepGEMM.git@v2.1.1.post3 --no-build-isolation
+```
+[vLLM Official Guide](https://docs.vllm.ai/projects/recipes/en/latest/Qwen/Qwen3.5.html)
+
+
+### 【vLLM Startup Command】
+<i>Note: When launching with TP=8, include `--enable-expert-parallel`; 
+otherwise the expert tensors wouldn’t be evenly sharded across GPU devices.</i>
+
+```
+export VLLM_USE_DEEP_GEMM=0
+export VLLM_USE_FLASHINFER_MOE_FP16=1
+export VLLM_USE_FLASHINFER_SAMPLER=0
+export OMP_NUM_THREADS=4
+
+vllm serve \
+    __YOUR_PATH__/QuantTrio/GLM-5-AWQ \
+    --served-model-name MY_MODEL \
+    --swap-space 16 \
+    --max-num-seqs 32 \
+    --max-model-len 32768  \
+    --gpu-memory-utilization 0.9 \
+    --tensor-parallel-size 8 \
+    --enable-expert-parallel \ 
+    --enable-auto-tool-choice \    
+    --tool-call-parser glm47 \
+    --reasoning-parser glm45 \
+    --speculative-config '{"method":"mtp","num_speculative_tokens":1}' \
+    --trust-remote-code \
+    --host 0.0.0.0 \
+    --port 8000
+```
+
+### 【Logs】
+```
+2026-02-26
+1. Initial commit
+```
+
+### 【Model Files】
+| File Size | Last Updated |
+|-----------|--------------|
+| `392 GiB` | `2026-02-26` |
+
+### 【Model Download】
+```python
+from huggingface_hub import snapshot_download
+snapshot_download('QuantTrio/GLM-5-AWQ', cache_dir="your_local_path")
+```
+
+### 【Overview】
+
+# GLM-5
+
+<div align="center">
+<img src=https://raw.githubusercontent.com/zai-org/GLM-5/refs/heads/main/resources/logo.svg width="15%"/>
+</div>
+<p align="center">
+    👋 Join our <a href="https://raw.githubusercontent.com/zai-org/GLM-5/refs/heads/main/resources/wechat.png" target="_blank">WeChat</a> or <a href="https://discord.gg/QR7SARHRxK" target="_blank">Discord</a> community.
+    <br>
+    📖 Check out the GLM-5 <a href="https://z.ai/blog/glm-5" target="_blank">technical blog</a>.
+    <br>
+    📍 Use GLM-5 API services on <a href="https://docs.z.ai/guides/llm/glm-5">Z.ai API Platform. </a>
+    <br>
+    👉 One click to <a href="https://chat.z.ai">GLM-5</a>.
+</p>
+
+## Introduction
+
+We are launching GLM-5, targeting complex systems engineering and long-horizon agentic tasks. Scaling is still one of the most important ways to improve the intelligence efficiency of Artificial General Intelligence (AGI). Compared to GLM-4.5, GLM-5 scales from 355B parameters (32B active) to 744B parameters (40B active), and increases pre-training data from 23T to 28.5T tokens. GLM-5 also integrates DeepSeek Sparse Attention (DSA), largely reducing deployment cost while preserving long-context capacity.
+
+Reinforcement learning aims to bridge the gap between competence and excellence in pre-trained models. However, deploying it at scale for LLMs is a challenge due to the RL training inefficiency. To this end, we developed [slime](https://github.com/THUDM/slime), a novel **asynchronous RL infrastructure** that substantially improves training throughput and efficiency, enabling more fine-grained post-training iterations. With advances in both pre-training and post-training, GLM-5 delivers significant improvement compared to GLM-4.7 across a wide range of academic benchmarks and achieves best-in-class performance among all open-source models in the world on reasoning, coding, and agentic tasks,  closing the gap with frontier models.
+
+## Benchmark
+
+|                                  | GLM-5                  | GLM-4.7   | DeepSeek-V3.2 | Kimi K2.5 | Claude Opus 4.5 | Gemini 3 Pro | GPT-5.2 (xhigh) |
+| -------------------------------- | ---------------------- | --------- | ------------- |-----------| --------------- | ------------ | --------------- |
+| HLE                              | 30.5                   | 24.8      | 25.1          | 31.5      | 28.4            | 37.2         | 35.4            |
+| HLE (w/ Tools)                   | 50.4                   | 42.8      | 40.8          | 51.8      | 43.4*           | 45.8*        | 45.5*           |
+| AIME 2026 I                      | 92.7                   | 92.9      | 92.7          | 92.5      | 93.3            | 90.6         | -               |
+| HMMT Nov. 2025                   | 96.9                   | 93.5      | 90.2          | 91.1      | 91.7            | 93.0         | 97.1            |
+| IMOAnswerBench                   | 82.5                   | 82.0      | 78.3          | 81.8      | 78.5            | 83.3         | 86.3            |
+| GPQA-Diamond                     | 86.0                   | 85.7      | 82.4          | 87.6      | 87.0            | 91.9         | 92.4            |
+| SWE-bench Verified               | 77.8                   | 73.8      | 73.1          | 76.8      | 80.9            | 76.2         | 80.0            |
+| SWE-bench Multilingual           | 73.3                   | 66.7      | 70.2          | 73.0      | 77.5            | 65.0         | 72.0            |
+| Terminal-Bench 2.0 (Terminus 2)  | 56.2 / 60.7 † | 41.0      | 39.3          | 50.8      | 59.3            | 54.2         | 54.0            |
+| Terminal-Bench 2.0 (Claude Code) | 56.2 / 61.1 †  | 32.8      | 46.4          | -         | 57.9            | -            | -               |
+| CyberGym                         | 43.2                   | 23.5      | 17.3          | 41.3      | 50.6            | 39.9         | -               |
+| BrowseComp                       | 62.0                   | 52.0      | 51.4          | 60.6      | 37.0            | 37.8         | -               |
+| BrowseComp (w/ Context Manage)   | 75.9                   | 67.5      | 67.6          | 74.9      | 67.8            | 59.2         | 65.8            |
+| BrowseComp-Zh                    | 72.7                   | 66.6      | 65.0          | 62.3      | 62.4            | 66.8         | 76.1            |
+| τ²-Bench                         | 89.7                   | 87.4      | 85.3          | 80.2      | 91.6            | 90.7         | 85.5            |
+| MCP-Atlas (Public Set)           | 67.8                   | 52.0      | 62.2          | 63.8      | 65.2            | 66.6         | 68.0            |
+| Tool-Decathlon                   | 38.0                   | 23.8      | 35.2          | 27.8      | 43.5            | 36.4         | 46.3            |
+| Vending Bench 2                  | $4,432.12              | $2,376.82 | $1,034.00     | $1,198.46 | $4,967.06       | $5,478.16    | $3,591.33       |
+
+> *: refers to their scores of full set.
+> 
+> †: A verified version of Terminal-Bench 2.0 that fixes some ambiguous instructions.
+See footnote for more evaluation details.
+
+### Footnote
+
+* **Humanity’s Last Exam (HLE) & other reasoning tasks**: We evaluate with a maximum generation length of 131,072 tokens (`temperature=1.0, top_p=0.95, max_new_tokens=131072`). By default, we report the text-only subset; results marked with * are from the full set. We use GPT-5.2 (medium) as the judge model. For HLE-with-tools, we use a maximum context length of 202,752 tokens.
+* **SWE-bench & SWE-bench Multilingual**: We run the SWE-bench suite with OpenHands using a tailored instruction prompt. Settings: `temperature=0.7, top_p=0.95, max_new_tokens=16384`, with a 200K context window.
+* **BrowserComp**: Without context management, we retain details from the most recent 5 turns. With context management, we use the same discard-all strategy as DeepSeek-v3.2 and Kimi K2.5.
+* **Terminal-Bench 2.0 (Terminus 2)**: We evaluate with the Terminus framework using `timeout=2h, temperature=0.7, top_p=1.0, max_new_tokens=8192`, with a 128K context window. Resource limits are capped at 16 CPUs and 32 GB RAM.
+* **Terminal-Bench 2.0 (Claude Code)**: We evaluate in Claude Code 2.1.14 (think mode, default effort) with `temperature=1.0, top_p=0.95, max_new_tokens=65536`. We remove wall-clock time limits due to generation speed, while preserving per-task CPU and memory constraints. Scores are averaged over 5 runs. We fix environment issues introduced by Claude Code and also report results on a verified Terminal-Bench 2.0 dataset that resolves ambiguous instructions (see: [https://huggingface.co/datasets/zai-org/terminal-bench-2-verified](https://huggingface.co/datasets/zai-org/terminal-bench-2-verified)).
+* **CyberGym**: We evaluate in Claude Code 2.1.18 (think mode, no web tools) with (`temperature=1.0, top_p=1.0, max_new_tokens=32000`) and a 250-minute timeout per task. Results are single-run Pass@1 over 1,507 tasks.
+* **MCP-Atlas**: All models are evaluated in think mode on the 500-task public subset with a 10-minute timeout per task. We use Gemini 3 Pro as the judge model.
+* **τ²-bench**: We add a small prompt adjustment in Retail and Telecom to avoid failures caused by premature user termination. For Airline, we apply the domain fixes proposed in the Claude Opus 4.5 system card.
+* **Vending Bench 2**: Runs are conducted independently by [Andon Labs](https://andonlabs.com/evals/vending-bench-2).
+
+
+## Serve GLM-5 Locally
+
+### Prepare environment
+
+vLLM, SGLang, and xLLM all support local deployment of GLM-5. A simple deployment guide is provided here.
+
++ vLLM
+
+    Using Docker as:
+
+    ```shell
+    docker pull vllm/vllm-openai:nightly 
+    ```
+
+    or using pip:
+
+    ```shell
+    pip install -U vllm --pre --index-url https://pypi.org/simple --extra-index-url https://wheels.vllm.ai/nightly
+    ```
+
+    then upgrade transformers:
+
+    ```
+    pip install git+https://github.com/huggingface/transformers.git
+    ```
+
++ SGLang
+
+    Using Docker as:
+    ```bash
+    docker pull lmsysorg/sglang:glm5-hopper # For Hopper GPU
+    docker pull lmsysorg/sglang:glm5-blackwell # For Blackwell GPU
+    ```
+
+### Deploy
+
++ vLLM
+
+    ```shell
+    vllm serve zai-org/GLM-5-FP8 \
+         --tensor-parallel-size 8 \
+         --gpu-memory-utilization 0.85 \
+         --speculative-config.method mtp \
+         --speculative-config.num_speculative_tokens 1 \
+         --tool-call-parser glm47 \
+         --reasoning-parser glm45 \
+         --enable-auto-tool-choice \
+         --served-model-name glm-5-fp8
+    ```
+
+    Check the [recipes](https://github.com/vllm-project/recipes/blob/main/GLM/GLM5.md) for more details.
+
++ SGLang
+
+    ```shell
+    python3 -m sglang.launch_server \
+      --model-path zai-org/GLM-5-FP8 \
+      --tp-size 8 \
+      --tool-call-parser glm47  \
+      --reasoning-parser glm45 \
+      --speculative-algorithm EAGLE \
+      --speculative-num-steps 3 \
+      --speculative-eagle-topk 1 \
+      --speculative-num-draft-tokens 4 \
+      --mem-fraction-static 0.85 \
+      --served-model-name glm-5-fp8
+    ```
+  
+    Check the [sglang cookbook](https://cookbook.sglang.io/autoregressive/GLM/GLM-5) for more details.
+
++ xLLM and other Ascend NPU
+
+    Please check the deployment guide [here](https://github.com/zai-org/GLM-5/blob/main/example/ascend.md).
+
+
+## Citation
+
+Our technical report is coming soon.

chat_template.jinja

@@ -0,0 +1,86 @@
+[gMASK]<sop>
+{%- if tools -%}
+<|system|>
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{% for tool in tools %}
+{{ tool | tojson(ensure_ascii=False) }}
+{% endfor %}
+</tools>
+
+For each function call, output the function name and arguments within the following XML format:
+<tool_call>{function-name}<arg_key>{arg-key-1}</arg_key><arg_value>{arg-value-1}</arg_value><arg_key>{arg-key-2}</arg_key><arg_value>{arg-value-2}</arg_value>...</tool_call>{%- endif -%}
+{%- macro visible_text(content) -%}
+    {%- if content is string -%}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping -%}
+        {%- for item in content -%}
+            {%- if item is mapping and item.type == 'text' -%}
+                {{- item.text }}
+            {%- elif item is string -%}
+                {{- item }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{- content }}
+    {%- endif -%}
+{%- endmacro -%}
+{%- set ns = namespace(last_user_index=-1) %}
+{%- for m in messages %}
+    {%- if m.role == 'user' %}
+        {% set ns.last_user_index = loop.index0 -%}
+    {%- endif %}
+{%- endfor %}
+{% for m in messages %}
+{%- if m.role == 'user' -%}<|user|>{{ visible_text(m.content) }}
+{%- elif m.role == 'assistant' -%}
+<|assistant|>
+{%- set reasoning_content = '' %}
+{%- set content = visible_text(m.content) %}
+{%- if m.reasoning_content is string %}
+    {%- set reasoning_content = m.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 %}
+{%- if ((clear_thinking is defined and not clear_thinking) or loop.index0 > ns.last_user_index) and reasoning_content -%}
+{{ '<think>' + reasoning_content.strip() +  '</think>'}}
+{%- else -%}
+{{ '</think>' }}
+{%- endif -%}
+{%- if content.strip() -%}
+{{ content.strip() }}
+{%- endif -%}
+{% if m.tool_calls %}
+{% for tc in m.tool_calls %}
+{%- if tc.function %}
+    {%- set tc = tc.function %}
+{%- endif %}
+{{- '<tool_call>' + tc.name -}}
+{% set _args = tc.arguments %}{% for k, v in _args.items() %}<arg_key>{{ k }}</arg_key><arg_value>{{ v | tojson(ensure_ascii=False) if v is not string else v }}</arg_value>{% endfor %}</tool_call>{% endfor %}
+{% endif %}
+{%- elif m.role == 'tool' -%}
+{%- if m.content is string -%}
+{%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+    {{- '<|observation|>' }}
+{%- endif %}
+{{- '<tool_response>' }}
+{{- m.content }}
+{{- '</tool_response>' }}
+{%- else -%}
+<|observation|>{% for tr in m.content %}
+<tool_response>{{ tr.output if tr.output is defined else tr }}</tool_response>{% endfor -%}
+{% endif -%}
+{%- elif m.role == 'system' -%}
+<|system|>{{ visible_text(m.content) }}
+{%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    <|assistant|>{{- '</think>' if (enable_thinking is defined and not enable_thinking) else '<think>' -}}
+{%- endif -%}

config.json

@@ -0,0 +1,75 @@
+{
+  "architectures": [
+    "GlmMoeDsaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "dtype": "bfloat16",
+  "eos_token_id": [
+    154820,
+    154827,
+    154829
+  ],
+  "ep_size": 1,
+  "first_k_dense_replace": 3,
+  "head_dim": 64,
+  "hidden_act": "silu",
+  "hidden_size": 6144,
+  "index_head_dim": 128,
+  "index_n_heads": 32,
+  "index_topk": 2048,
+  "indexer_rope_interleave": true,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "kv_lora_rank": 512,
+  "max_position_embeddings": 202752,
+  "model_type": "glm_moe_dsa",
+  "moe_intermediate_size": 2048,
+  "moe_layer_freq": 1,
+  "n_group": 1,
+  "n_routed_experts": 205,
+  "n_shared_experts": 1,
+  "name_or_path": "tclf90/GLM-5-AWQ",
+  "norm_topk_prob": true,
+  "num_attention_heads": 64,
+  "num_experts_per_tok": 8,
+  "num_hidden_layers": 78,
+  "num_key_value_heads": 64,
+  "num_nextn_predict_layers": 1,
+  "pad_token_id": 154820,
+  "pretraining_tp": 1,
+  "q_lora_rank": 2048,
+  "qk_head_dim": 256,
+  "qk_nope_head_dim": 192,
+  "qk_rope_head_dim": 64,
+  "quantization_config": {
+    "bits": 4,
+    "group_size": 128,
+    "modules_to_not_convert": [
+      "self_attn",
+      "shared_expert",
+      "mlp.gate",
+      "model.layers.0.",
+      "model.layers.1.",
+      "model.layers.2."
+    ],
+    "quant_method": "awq",
+    "version": "gemm",
+    "zero_point": true
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_interleave": true,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "routed_scaling_factor": 2.5,
+  "scoring_func": "sigmoid",
+  "tie_word_embeddings": false,
+  "topk_group": 1,
+  "topk_method": "noaux_tc",
+  "transformers_version": "5.0.2.dev0",
+  "use_cache": true,
+  "v_head_dim": 256,
+  "vocab_size": 154880
+}

expert_prune_report.json

@@ -0,0 +1,83 @@
+{
+  "mode": "physical",
+  "new_num_experts": 205,
+  "old_num_experts": 256,
+  "per_layer_keep_counts": {
+    "10": 205,
+    "11": 205,
+    "12": 205,
+    "13": 205,
+    "14": 205,
+    "15": 205,
+    "16": 205,
+    "17": 205,
+    "18": 205,
+    "19": 205,
+    "20": 205,
+    "21": 205,
+    "22": 205,
+    "23": 205,
+    "24": 205,
+    "25": 205,
+    "26": 205,
+    "27": 205,
+    "28": 205,
+    "29": 205,
+    "3": 205,
+    "30": 205,
+    "31": 205,
+    "32": 205,
+    "33": 205,
+    "34": 205,
+    "35": 205,
+    "36": 205,
+    "37": 205,
+    "38": 205,
+    "39": 205,
+    "4": 205,
+    "40": 205,
+    "41": 205,
+    "42": 205,
+    "43": 205,
+    "44": 205,
+    "45": 205,
+    "46": 205,
+    "47": 205,
+    "48": 205,
+    "49": 205,
+    "5": 205,
+    "50": 205,
+    "51": 205,
+    "52": 205,
+    "53": 205,
+    "54": 205,
+    "55": 205,
+    "56": 205,
+    "57": 205,
+    "58": 205,
+    "59": 205,
+    "6": 205,
+    "60": 205,
+    "61": 205,
+    "62": 205,
+    "63": 205,
+    "64": 205,
+    "65": 205,
+    "66": 205,
+    "67": 205,
+    "68": 205,
+    "69": 205,
+    "7": 205,
+    "70": 205,
+    "71": 205,
+    "72": 205,
+    "73": 205,
+    "74": 205,
+    "75": 205,
+    "76": 205,
+    "77": 205,
+    "8": 205,
+    "9": 205
+  },
+  "warnings": []
+}

generation_config.json

@@ -0,0 +1,12 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    154820,
+    154827,
+    154829
+  ],
+  "pad_token_id": 154820,
+  "temperature": 1.0,
+  "top_p": 0.95,
+  "transformers_version": "5.0.2.dev0"
+}

glm47_moe_tool_parser_fixed.py

@@ -0,0 +1,532 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+"""
+GLM-4 Tool Call Parser with incremental string streaming support.
+
+This parser fixes the streaming issue reported in Issue #32829 where long string
+parameters (e.g., file content with 4000+ characters of code) are buffered until
+complete, causing multi-second delays before the user sees any content.
+
+The fix streams string values incrementally as they arrive, providing a true
+streaming experience for long content.
+"""
+
+import ast
+import json
+from collections.abc import Sequence
+from typing import Any
+
+import regex as re
+
+from vllm.entrypoints.chat_utils import make_tool_call_id
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+    ChatCompletionToolsParam,
+)
+from vllm.entrypoints.openai.engine.protocol import (
+    DeltaFunctionCall,
+    DeltaMessage,
+    DeltaToolCall,
+    ExtractedToolCallInformation,
+    FunctionCall,
+    ToolCall,
+)
+from vllm.logger import init_logger
+from vllm.tokenizers import TokenizerLike
+from vllm.tool_parsers.abstract_tool_parser import (
+    ToolParser,
+    ToolParserManager,
+)
+
+logger = init_logger(__name__)
+
+
+@ToolParserManager.register_module("glm47_fixed")
+class Glm47MoeModelToolParser(ToolParser):
+    """Tool parser for GLM-4 models with incremental string streaming.
+
+    This parser emits tool-call deltas incrementally as arguments arrive.
+    For string-type parameters, content is streamed character-by-character
+    rather than waiting for the complete </arg_value> tag.
+    """
+
+    def __init__(self, tokenizer: TokenizerLike):
+        super().__init__(tokenizer)
+        # Stateful streaming fields
+        self.current_tool_name_sent: bool = False
+        self.prev_tool_call_arr: list[dict[str, Any]] = []
+        self.current_tool_id: int = -1
+        self.streamed_args_for_tool: list[str] = []
+
+        self.tool_call_start_token: str = "<tool_call>"
+        self.tool_call_end_token: str = "</tool_call>"
+        self.arg_key_start: str = "<arg_key>"
+        self.arg_key_end: str = "</arg_key>"
+        self.arg_val_start: str = "<arg_value>"
+        self.arg_val_end: str = "</arg_value>"
+
+        self.tool_calls_start_token = self.tool_call_start_token
+
+        self.func_call_regex = re.compile(r"<tool_call>.*?</tool_call>", re.DOTALL)
+
+        # GLM-4.7 format: <tool_call>func_name[<arg_key>...]*</tool_call>
+        # The function name can be followed by a newline, whitespace, or
+        # directly by <arg_key> tags (no separator).  The arg section is
+        # optional so that zero-argument calls are supported.
+        self.func_detail_regex = re.compile(
+            r"<tool_call>\s*(\S+?)\s*(<arg_key>.*)?</tool_call>", re.DOTALL
+        )
+        self.func_arg_regex = re.compile(
+            r"<arg_key>(.*?)</arg_key>\s*<arg_value>(.*?)</arg_value>",
+            re.DOTALL,
+        )
+
+        if not self.model_tokenizer:
+            raise ValueError(
+                "The model tokenizer must be passed to the ToolParser "
+                "constructor during construction."
+            )
+
+        self.tool_call_start_token_id = self.vocab.get(self.tool_call_start_token)
+        self.tool_call_end_token_id = self.vocab.get(self.tool_call_end_token)
+        self._buffer: str = ""
+
+        # Streaming state for incremental tool-call streaming
+        self._in_tool_call: bool = False
+        self._current_tool_name: str | None = None
+        self._pending_key: str | None = None
+        self._streaming_string_value: bool = False
+        self._tool_call_ids: list[str] = []
+        self._args_started: list[bool] = []
+        self._args_closed: list[bool] = []
+        self._seen_keys: list[set[str]] = []
+
+    @staticmethod
+    def _deserialize(value: str) -> Any:
+        try:
+            return json.loads(value)
+        except json.JSONDecodeError:
+            pass
+
+        try:
+            return ast.literal_eval(value)
+        except (ValueError, SyntaxError):
+            pass
+
+        return value
+
+    @staticmethod
+    def _json_escape_string_content(s: str) -> str:
+        """JSON-escape string content for incremental streaming.
+
+        This escapes the content that goes INSIDE a JSON string (between quotes),
+        not including the surrounding quotes themselves.
+        """
+        if not s:
+            return ""
+        return json.dumps(s, ensure_ascii=False)[1:-1]
+
+    @staticmethod
+    def _is_string_type(
+        tool_name: str,
+        arg_name: str,
+        tools: list[ChatCompletionToolsParam] | None,
+    ) -> bool:
+        if tools is None:
+            return False
+        for tool in tools:
+            if tool.function.name != tool_name:
+                continue
+            if tool.function.parameters is None:
+                return False
+            arg_type = (
+                tool.function.parameters.get("properties", {})
+                .get(arg_name, {})
+                .get("type", None)
+            )
+            return arg_type == "string"
+        logger.debug("No tool named '%s'.", tool_name)
+        return False
+
+    @staticmethod
+    def _tools_enabled(request: ChatCompletionRequest) -> bool:
+        """Return whether tool parsing should be applied for this request."""
+        try:
+            tools = getattr(request, "tools", None)
+            tool_choice = getattr(request, "tool_choice", None)
+            return bool(tools) and tool_choice != "none"
+        except Exception:
+            logger.exception("Failed to determine if tools are enabled.")
+            return False
+
+    def adjust_request(self, request: ChatCompletionRequest) -> ChatCompletionRequest:
+        """Adjust request parameters for tool call token handling."""
+        request = super().adjust_request(request)
+        if request.tools and request.tool_choice != "none":
+            # Ensure tool call tokens (<tool_call>, </tool_call>) are not skipped
+            # during decoding. Even though they are not marked as special tokens,
+            # setting skip_special_tokens=False ensures proper handling in
+            # transformers 5.x where decoding behavior may have changed.
+            request.skip_special_tokens = False
+        return request
+
+    def extract_tool_calls(
+        self,
+        model_output: str,
+        request: ChatCompletionRequest,
+    ) -> ExtractedToolCallInformation:
+        matched_tool_calls = self.func_call_regex.findall(model_output)
+        logger.debug("model_output: %s", model_output)
+        try:
+            tool_calls: list[ToolCall] = []
+            for match in matched_tool_calls:
+                tc_detail = self.func_detail_regex.search(match)
+                if not tc_detail:
+                    logger.warning(
+                        "Failed to parse tool call details from: %s",
+                        match,
+                    )
+                    continue
+                tc_name = tc_detail.group(1).strip()
+                tc_args = tc_detail.group(2)
+                pairs = self.func_arg_regex.findall(tc_args) if tc_args else []
+                arg_dct: dict[str, Any] = {}
+                for key, value in pairs:
+                    arg_key = key.strip()
+                    arg_val = value.strip()
+                    if not self._is_string_type(tc_name, arg_key, request.tools):
+                        arg_val = self._deserialize(arg_val)
+                    logger.debug("arg_key = %s, arg_val = %s", arg_key, arg_val)
+                    arg_dct[arg_key] = arg_val
+                tool_calls.append(
+                    ToolCall(
+                        type="function",
+                        function=FunctionCall(
+                            name=tc_name,
+                            arguments=json.dumps(arg_dct, ensure_ascii=False),
+                        ),
+                    )
+                )
+        except Exception:
+            logger.exception("Failed to extract tool call spec")
+            return ExtractedToolCallInformation(
+                tools_called=False, tool_calls=[], content=model_output
+            )
+        else:
+            if len(tool_calls) > 0:
+                content: str | None = model_output[
+                    : model_output.find(self.tool_calls_start_token)
+                ]
+                # Normalize empty/whitespace-only content to None
+                if not content or not content.strip():
+                    content = None
+                return ExtractedToolCallInformation(
+                    tools_called=True, tool_calls=tool_calls, content=content
+                )
+            return ExtractedToolCallInformation(
+                tools_called=False, tool_calls=[], content=model_output
+            )
+
+    def extract_tool_calls_streaming(
+        self,
+        previous_text: str,
+        current_text: str,
+        delta_text: str,
+        previous_token_ids: Sequence[int],
+        current_token_ids: Sequence[int],
+        delta_token_ids: Sequence[int],
+        request: ChatCompletionRequest,
+    ) -> DeltaMessage | None:
+        if not self._tools_enabled(request):
+            return DeltaMessage(content=delta_text) if delta_text else None
+
+        self._buffer += delta_text
+
+        while True:
+            if not self._in_tool_call:
+                start_idx = self._buffer.find(self.tool_call_start_token)
+                if start_idx == -1:
+                    # Check for partial start token at end of buffer
+                    for i in range(1, len(self.tool_call_start_token)):
+                        if self._buffer.endswith(self.tool_call_start_token[:i]):
+                            out = self._buffer[:-i]
+                            self._buffer = self._buffer[-i:]
+                            return DeltaMessage(content=out) if out else None
+                    out = self._buffer
+                    self._buffer = ""
+                    return DeltaMessage(content=out) if out else None
+
+                if start_idx > 0:
+                    out = self._buffer[:start_idx]
+                    self._buffer = self._buffer[start_idx:]
+                    return DeltaMessage(content=out) if out else None
+
+                self._buffer = self._buffer[len(self.tool_call_start_token) :]
+                self._begin_tool_call()
+                continue
+
+            # Parse tool name first
+            if not self.current_tool_name_sent:
+                nl = self._buffer.find("\n")
+                ak = self._buffer.find(self.arg_key_start)
+                end = self._buffer.find(self.tool_call_end_token)
+                candidates = [i for i in [nl, ak, end] if i != -1]
+                if not candidates:
+                    return None
+                cut = min(candidates)
+                tool_name = self._buffer[:cut].strip()
+                if tool_name == "" and cut == end:
+                    # Handle empty tool call like `<tool_call></tool_call>`.
+                    # Consume the tokens and reset state to avoid infinite loop.
+                    self._buffer = self._buffer[end + len(self.tool_call_end_token) :]
+                    self._finish_tool_call()
+                    self._revert_last_tool_call_state()
+                    continue
+
+                if cut == nl:
+                    self._buffer = self._buffer[nl + 1 :]
+                else:
+                    self._buffer = self._buffer[cut:]
+
+                self._current_tool_name = tool_name
+                self.current_tool_name_sent = True
+                return self._emit_tool_name_delta(tool_name)
+
+            assert self._current_tool_name is not None
+
+            # Handle incremental string value streaming
+            if self._streaming_string_value:
+                val_end = self._buffer.find(self.arg_val_end)
+                if val_end != -1:
+                    raw_content = self._buffer[:val_end]
+                    self._buffer = self._buffer[val_end + len(self.arg_val_end) :]
+                    self._streaming_string_value = False
+                    self._pending_key = None
+
+                    escaped = self._json_escape_string_content(raw_content)
+                    frag = escaped + '"'
+                    self.streamed_args_for_tool[self.current_tool_id] += frag
+                    return self._emit_tool_args_delta(frag)
+                else:
+                    # Check for partial </arg_value> at end
+                    safe_len = len(self._buffer)
+                    for i in range(1, len(self.arg_val_end)):
+                        if self._buffer.endswith(self.arg_val_end[:i]):
+                            safe_len = len(self._buffer) - i
+                            break
+
+                    if safe_len > 0:
+                        to_emit = self._buffer[:safe_len]
+                        self._buffer = self._buffer[safe_len:]
+                        escaped = self._json_escape_string_content(to_emit)
+                        if escaped:
+                            self.streamed_args_for_tool[self.current_tool_id] += escaped
+                            return self._emit_tool_args_delta(escaped)
+                    return None
+
+            # If we have a pending key, parse its value
+            if self._pending_key is not None:
+                val_pos = self._buffer.find(self.arg_val_start)
+                if val_pos == -1:
+                    return None
+                if val_pos > 0:
+                    self._buffer = self._buffer[val_pos:]
+
+                key = (self._pending_key or "").strip()
+
+                is_string = self._is_string_type(
+                    self._current_tool_name, key, request.tools
+                )
+
+                if is_string:
+                    # String type: stream incrementally
+                    self._buffer = self._buffer[len(self.arg_val_start) :]
+
+                    if key in self._seen_keys[self.current_tool_id]:
+                        self._pending_key = None
+                        continue
+
+                    self._seen_keys[self.current_tool_id].add(key)
+                    key_json = json.dumps(key, ensure_ascii=False)
+
+                    if not self._args_started[self.current_tool_id]:
+                        frag = "{" + key_json + ': "'
+                        self._args_started[self.current_tool_id] = True
+                    else:
+                        frag = ", " + key_json + ': "'
+
+                    self.streamed_args_for_tool[self.current_tool_id] += frag
+                    self._streaming_string_value = True
+                    return self._emit_tool_args_delta(frag)
+                else:
+                    # Non-string type: wait for complete value
+                    val_end = self._buffer.find(self.arg_val_end)
+                    if val_end == -1:
+                        return None
+
+                    raw_val = self._buffer[len(self.arg_val_start) : val_end].strip()
+                    self._buffer = self._buffer[val_end + len(self.arg_val_end) :]
+                    self._pending_key = None
+
+                    frag_or_none = self._append_arg_fragment(key=key, raw_val=raw_val)
+                    if frag_or_none:
+                        return self._emit_tool_args_delta(frag_or_none)
+                    continue
+
+            # Parse next arg or close
+            end_pos = self._buffer.find(self.tool_call_end_token)
+            key_pos = self._buffer.find(self.arg_key_start)
+            if end_pos != -1 and (key_pos == -1 or end_pos < key_pos):
+                self._buffer = self._buffer[end_pos + len(self.tool_call_end_token) :]
+                frag_or_none = self._close_args_if_needed()
+                # Finalize prev_tool_call_arr with complete parsed arguments
+                if self._current_tool_name:
+                    try:
+                        full_args_str = self.streamed_args_for_tool[
+                            self.current_tool_id
+                        ]
+                        json.loads(full_args_str)
+                        self.prev_tool_call_arr[self.current_tool_id] = {
+                            "name": self._current_tool_name,
+                            "arguments": full_args_str,
+                        }
+                    except (json.JSONDecodeError, IndexError) as e:
+                        logger.warning(
+                            "Failed to finalize tool call state for tool %d: %s",
+                            self.current_tool_id,
+                            e,
+                        )
+                self._finish_tool_call()
+                return (
+                    self._emit_tool_args_delta(frag_or_none) if frag_or_none else None
+                )
+
+            if key_pos == -1:
+                return None
+            if key_pos > 0:
+                self._buffer = self._buffer[key_pos:]
+            key_end = self._buffer.find(self.arg_key_end)
+            if key_end == -1:
+                return None
+            key = self._buffer[len(self.arg_key_start) : key_end]
+            self._buffer = self._buffer[key_end + len(self.arg_key_end) :]
+            self._pending_key = key
+            continue
+
+    def _ensure_tool_state(self) -> None:
+        while len(self._tool_call_ids) <= self.current_tool_id:
+            self._tool_call_ids.append(
+                make_tool_call_id(id_type="random", func_name=None, idx=None)
+            )
+        while len(self.streamed_args_for_tool) <= self.current_tool_id:
+            self.streamed_args_for_tool.append("")
+        while len(self.prev_tool_call_arr) <= self.current_tool_id:
+            self.prev_tool_call_arr.append({})
+        while len(self._args_started) <= self.current_tool_id:
+            self._args_started.append(False)
+        while len(self._args_closed) <= self.current_tool_id:
+            self._args_closed.append(False)
+        while len(self._seen_keys) <= self.current_tool_id:
+            self._seen_keys.append(set())
+
+    def _begin_tool_call(self) -> None:
+        if self.current_tool_id == -1:
+            self.current_tool_id = 0
+        else:
+            self.current_tool_id += 1
+        self._ensure_tool_state()
+        self.current_tool_name_sent = False
+        self._current_tool_name = None
+        self._pending_key = None
+        self._streaming_string_value = False
+        self._in_tool_call = True
+
+    def _finish_tool_call(self) -> None:
+        self._in_tool_call = False
+        self._current_tool_name = None
+        self._pending_key = None
+        self._streaming_string_value = False
+
+    def _revert_last_tool_call_state(self) -> None:
+        """Revert the state allocation for the last tool call."""
+        if self.current_tool_id < 0:
+            return
+        self._tool_call_ids.pop()
+        self.streamed_args_for_tool.pop()
+        self.prev_tool_call_arr.pop()
+        self._args_started.pop()
+        self._args_closed.pop()
+        self._seen_keys.pop()
+        self.current_tool_id -= 1
+
+    def _emit_tool_name_delta(self, tool_name: str) -> DeltaMessage:
+        self.prev_tool_call_arr[self.current_tool_id] = {
+            "name": self._current_tool_name,
+            "arguments": {},
+        }
+        return DeltaMessage(
+            tool_calls=[
+                DeltaToolCall(
+                    index=self.current_tool_id,
+                    id=self._tool_call_ids[self.current_tool_id],
+                    type="function",
+                    function=DeltaFunctionCall(
+                        name=tool_name,
+                        arguments="",
+                    ).model_dump(exclude_none=True),
+                )
+            ]
+        )
+
+    def _emit_tool_args_delta(self, fragment: str) -> DeltaMessage:
+        return DeltaMessage(
+            tool_calls=[
+                DeltaToolCall(
+                    index=self.current_tool_id,
+                    function=DeltaFunctionCall(arguments=fragment).model_dump(
+                        exclude_none=True
+                    ),
+                )
+            ]
+        )
+
+    def _append_arg_fragment(
+        self,
+        *,
+        key: str,
+        raw_val: str,
+    ) -> str | None:
+        key = key.strip()
+        if not key:
+            return None
+        if key in self._seen_keys[self.current_tool_id]:
+            return None
+
+        # This function is only called for non-string types (already checked
+        # by _is_string_type in the caller), so we always deserialize.
+        val_obj: Any = self._deserialize(raw_val)
+
+        key_json = json.dumps(key, ensure_ascii=False)
+        val_json = json.dumps(val_obj, ensure_ascii=False)
+
+        if not self._args_started[self.current_tool_id]:
+            fragment = "{" + key_json + ": " + val_json
+            self._args_started[self.current_tool_id] = True
+        else:
+            fragment = "," + key_json + ": " + val_json
+
+        self._seen_keys[self.current_tool_id].add(key)
+        self.streamed_args_for_tool[self.current_tool_id] += fragment
+        return fragment
+
+    def _close_args_if_needed(self) -> str | None:
+        if self._args_closed[self.current_tool_id]:
+            return None
+        self._args_closed[self.current_tool_id] = True
+        if not self._args_started[self.current_tool_id]:
+            fragment = "{}"
+            self.streamed_args_for_tool[self.current_tool_id] = fragment
+        else:
+            fragment = "}"
+            self.streamed_args_for_tool[self.current_tool_id] += fragment
+        return fragment

model-00001-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bfa7b1b3d65b6628c54802bf51b4a0e6c4222978746e0cf6e2935d2ba2f36a3
+size 2993793440

model-00002-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4188dbdbfec85e375b368d7c1ca6df4eadb9e27373448258b05c2b9840408e8
+size 1993962224

model-00003-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52fe64ac77031721f717414465cfd15fb04ebcffded0538d5bfceba7b4b9050d
+size 2684816444

model-00004-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d82772da54764bdc80fc5476e5dda449a59964d1a77d1cfe7c34c6f8fe22b6e
+size 2307766008

model-00005-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c6ad662939d781548a7e0e7b60e53010a571974577406a665bc63bba2329fe4
+size 2325249572

model-00006-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74f0faa66fa323b0865aeec6bc1cd863bd89fe3c48f18fbfb956d4cdced1f109
+size 2667332872

model-00007-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5e47fa6153c18fbd1d62c0f9ee4228abd5d584de23eb3041c7d50c96d5c8ee4
+size 1998370612

model-00008-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:871ef6b1801b879e14404e1f92e27a6652852ba1a0d67e5b6c58f5f5170f46a6
+size 2940355204

model-00009-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d31c474890ab4091daa28b98785591ab84c0c4f47be6d7a7f119bc9c360bcc0c
+size 1999730088

model-00010-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4ef9952b191c5e3dfc28dbb82424875974134b755539b9fd62c73dfad917405
+size 2998817148

model-00011-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b591800f6c5c42f401b9c010b10fe09752597f023762a1b5d212ece3d441bb0
+size 1993962288

model-00012-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4b717ba2a951bf06bfb4c75c18d770b9cdc9c21724a29cad6167b80fd0bb9c3
+size 2887481476

model-00013-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:061bf24bc616de133fef2940598a9c35d46d6f87faad83d52ce76d371aafc291
+size 2105101056

model-00014-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3aa9027d1d1465fb43b0563c8281e7c0f1a969476a6fbb667cfd80215b14e78
+size 2625978732

model-00015-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a820a4068cdea35ef7364efc01e1dd32a9d32129bf027a11b3ed3a27a5708b19
+size 2366605248

model-00016-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e936b0df2d6cf9cd03265c007aaaa5aebbbdccb1dd71b9ed648465f2dbd01255
+size 2135660532

model-00017-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2862ba86b3f6bd30193f1d73ebc351e73a0ab599c89cc77d2627c9c2d4302e4d
+size 2856924024

model-00018-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5911ebf7cda3cf88fd7b87d0f530c43893cce08511f6c58cf5b6ac939dd4852
+size 1998371452

model-00019-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5e108b4c13ae03bd861e42785f81d02f3ea6198b99d693d987f0c42d640e64c
+size 2957650604

model-00020-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cdc6b1d14645b6b1b0d396910f4f76369a4ad7174554ce5766ce049cce2136e
+size 1995708680

model-00021-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d901f3ae485917cd900bb5b0c3c707eaba94dce96a5dfabf7aaf25ce50e23cc5
+size 2998621612

model-00022-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7fd915528301d9881dad67edffd93dd9679b3529bf686ebc4d23875bce70df6
+size 1993963184

model-00023-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d2b08a3c334e90d33b5685cda8b9922018515c194fcdc2a7c6810eb70728577
+size 2684817444

model-00024-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3a8c91cdb474c7e9654986eedfb18a232281264597c3eac59f1ac8e4b568b73
+size 2307767112

model-00025-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42b15f859a103ffe7277d14ec4033252aedf26cb30e2c5eacad214649a722df1
+size 2429851804

model-00026-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b8a5f54e11d9a1cf1424e0c867fb4cc3a13d217961c2e29e014940381589dce
+size 2562732760

model-00027-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8756bfc840b3197a62ab1696000dcd44f0c88a18844f246c5b8fd8f21290bc99
+size 1998371308

model-00028-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54957a027609361dd18262bc150dee9cc2edef1f6d1406edd9615caba4e51c51
+size 2998278716

model-00029-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b932b1a0902da4143ae1edfaac2c735b49d756cf1ddfd01b4f1710c5b9b547be
+size 1994306080

model-00031-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89c97694b76e73021ffc12926593bb104f93e0f5fb6bca86d7b7f6b0723241af
+size 1993963208

model-00125-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ebe5e018db4720770256d32c8ddc59c120201bdfb32e1e0c3dd9fad197c689d
+size 2684817444

model-00127-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ecb23a8bebd5dafb951fcb5733b14749e0140485ee80abaf356d6e6c9781e68
+size 2429851804

model-00129-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a2a740cbaad266c8e7f5ee80f386384cb5cb15eb44664d3dcb2bb50f9eae183
+size 1998371308

model-00130-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3040d373f6cb968f32193ca7933c1fd89c840ffe72453b140bd80ca0d9aeba1a
+size 2998278716

model-00133-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d993b3f240ae7c37e085779e3f747511b1f87d7bb253984e7bcdc38e9c829533
+size 1993963208

model-00135-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f2e7fd5e9a2d67b515548375187c985e2e089c524861da6f167d7f7f1adb278
+size 2046263768

model-00136-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d8fd824978a759fb2d3dd220cc3abb84f035e27a9f3107e02cfed886e1b080c
+size 2625979284

model-00141-of-00141.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcae2b399560f462a345a927232d4ace1697c1b981accef5cf8504ac66c55baf
+size 2054210160

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19e773648cb4e65de8660ea6365e10acca112d42a854923df93db4a6f333a82d
+size 20217442

tokenizer_config.json

@@ -0,0 +1,33 @@
+{
+  "backend": "tokenizers",
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": [
+    "<|endoftext|>",
+    "[MASK]",
+    "[gMASK]",
+    "[sMASK]",
+    "<sop>",
+    "<eop>",
+    "<|system|>",
+    "<|user|>",
+    "<|assistant|>",
+    "<|observation|>",
+    "<|begin_of_image|>",
+    "<|end_of_image|>",
+    "<|begin_of_video|>",
+    "<|end_of_video|>",
+    "<|begin_of_audio|>",
+    "<|end_of_audio|>",
+    "<|begin_of_transcription|>",
+    "<|end_of_transcription|>"
+  ],
+  "is_local": true,
+  "model_max_length": 202752,
+  "model_specific_special_tokens": {},
+  "pad_token": "<|endoftext|>",
+  "padding_side": "left",
+  "remove_space": false,
+  "tokenizer_class": "TokenizersBackend"
+}