Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ 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

README.md

@@ -0,0 +1,124 @@
+---
+license: other
+license_name: minimax-m2.7
+license_link: https://huggingface.co/MiniMaxAI/MiniMax-M2.7/blob/main/LICENSE
+base_model: MiniMaxAI/MiniMax-M2.7
+base_model_relation: quantized
+tags:
+  - prismaquant
+  - mixed-precision
+  - quantization
+  - nvfp4
+  - fp8
+  - moe
+  - vllm
+  - compressed-tensors
+library_name: vllm
+inference: false
+pipeline_tag: text-generation
+---
+
+# MiniMax-M2.7 — PrismaQuant 3.20 bpp (vLLM)
+
+Mixed-precision quantization of [MiniMax-M2.7](https://huggingface.co/MiniMaxAI/MiniMax-M2.7) using [prismaquant](https://github.com/RobTand/prismaquant). The artifact is a standard `compressed-tensors` checkpoint that vLLM serves natively — **no patches, no custom kernels, no forked runtime**.
+
+This is a **single DGX Spark target**: the model fits in 121 GB unified memory with KV cache headroom for 32k context.
+
+## Headline numbers
+
+| Metric | Value |
+|---|---|
+| Source model | MiniMax-M2.7 (228 B params, 256-expert MoE) |
+| Disk size | **90 GB** (-58% vs FP8 source ~215 GB; -73% vs BF16 ~456 GB) |
+| Achieved bpp | 3.20 |
+| Format mix | 30,780 NVFP4 + 2,204 FP8_SOURCE Linears |
+| Experts pruned | 4,960 of 15,872 (31 %), via REAP saliency + joint quant-prune knapsack |
+| MoE config | num_local_experts = 176 (down from 256), top_k = 8 |
+| vLLM patches required | **0** |
+
+## Quality
+
+Spot-checked on coherent generations across agentic / math / coding prompts at temperature 0:
+
+| Prompt | Result |
+|---|---|
+| Multi-segment train problem (math) | Step-by-step reasoning, exact answer 240 mi / 68.571 mph |
+| Python `is_palindrome` | Clean, correct |
+| Python `quicksort` | Clean, correct |
+| Python `binary_search` | Clean, correct |
+| Python `longest_substring_without_repeat` | Clean, sliding-window correct |
+| Python `merge_two_lists` (linked list) | Clean, correct |
+| Python `fibonacci` | Clean iterative impl with worked example |
+| Rust `Point::distance` | Clean, uses `.hypot()` (numerically stable) |
+| SQL top-5 customers by 2024 volume | Clean SQL, proper date-range filter |
+| Tool calling (function-call API) | Clean `{"city": "Paris"}` JSON emission |
+| Reasoning tags via `<think>` | Captured by `--reasoning-parser minimax_m2` |
+
+Formal benchmarks (MMLU, GSM8K, HumanEval) deferred to a follow-up. The artifact is positioned as a **fits-on-Spark, serves-coherently** working release; rigorous benchmark numbers are next.
+
+## Throughput
+
+| Metric | Value |
+|---|---|
+| Decode tok/s (single-stream, T0, ctx 128) | **~14 tok/s** |
+| Prior NVINT3 baseline on same hardware | 3.78 tok/s |
+| Speedup | **+265 %** |
+
+Measured against the same DGX Spark host (GB10) used for the prior NVINT3 release. The NVFP4 + FP8 mix avoids NVINT-load expansion that pushed earlier 3.25 bpp artifacts to 116 GiB at runtime; this artifact's runtime ≈ disk ≈ 90 GB.
+
+## Format mix
+
+```
+NVFP4       : 30,780 Linears (94.5 %)  — experts + most attention projections
+FP8_SOURCE  :  2,204 Linears  (6.7 %)  — passthrough of natively-FP8 source weights
+BF16        :     62 routers (router output dim shrunk to kept-experts only)
+PRUNED      : 14,880 Linear slots (4,960 experts × 3 weights, dropped per REAP)
+```
+
+## How it was produced
+
+prismaquant's allocator solves a multi-choice knapsack over per-Linear `Δloss ≈ 0.5 · H_trace · MSE_W`, where `H_trace` is the empirical Fisher diagonal trace from a streaming probe and `MSE_W` is the measured per-format round-trip error. Joint expert pruning composes into the same knapsack: each MoE choice is `(format, dropped_expert_ids)` priced together.
+
+For this release:
+
+- **Calibration**: 32 chunks × 4 samples × 2048 seqlen ≈ 262 k tokens from `cal-mix-v1` (multi-domain agentic + math + coding mix).
+- **Probe**: streaming layer-by-layer Fisher capture (incremental_probe).
+- **Cost step**: per-(Linear, format) NVFP4 / FP8 / MXFP8 / BF16 round-trip MSE.
+- **Allocator**: target-bits = 3.20 (knee within the user-specified 90-95 GB band).
+- **Export**: native compressed-tensors with GPTQ + per-group scale-sweep activation-aware refinement on every NVFP4 Linear.
+
+## Serving
+
+```bash
+vllm serve <this-repo> \
+  --quantization compressed-tensors \
+  --trust-remote-code \
+  --max-model-len 32768 \
+  --max-num-seqs 4 \
+  --gpu-memory-utilization 0.85 \
+  --kv-cache-dtype fp8 \
+  --enable-prefix-caching \
+  --enable-auto-tool-choice \
+  --tool-call-parser minimax_m2 \
+  --reasoning-parser minimax_m2
+```
+
+Recommended: set `PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True` on UMA hardware (DGX Spark) to keep the cuda allocator from hoarding freed blocks.
+
+## Limitations / caveats
+
+- **Calibration scale**: 262 k tokens is moderate. Heavy reasoning-chain or long-context tasks may benefit from a re-export against more diverse calibration data.
+- **No formal benchmarks yet**: MMLU / GSM8K / HumanEval pending. The headline result is "fits on Spark + coherent generations on representative prompts".
+- **MTP heads not present**: MiniMax-M2 has no MTP head (unlike Qwen3.5/3.6). No speculative decoding accelerator to ship alongside.
+- **Pruned experts are gone**: 31 % of experts dropped per REAP saliency. Tasks heavily dependent on those specific experts could see degradation. Empirical probes show none on agentic/math/coding prompts.
+
+## Acknowledgements
+
+- [MiniMaxAI](https://huggingface.co/MiniMaxAI) for the source model.
+- [vLLM](https://github.com/vllm-project/vllm) for the compressed-tensors serving stack.
+- [REAP](https://arxiv.org/abs/2410.21271) for the per-expert dropout-loss saliency formulation.
+- HAQ / HAWQ / GPTQ / AutoRound / SmoothQuant for the mixed-precision allocation foundations.
+
+## License
+
+Inherits the MiniMax-M2.7 license from the source model. See [base model card](https://huggingface.co/MiniMaxAI/MiniMax-M2.7) for terms.

chat_template.jinja

@@ -0,0 +1,159 @@
+{# ----------‑‑‑ special token variables ‑‑‑---------- #}
+{%- set toolcall_begin_token   = '<minimax:tool_call>'         -%}
+{%- set toolcall_end_token     = '</minimax:tool_call>'        -%}
+{#- Tool Rendering Functions ============================================== -#}
+{%- macro render_tool_namespace(namespace_name, tool_list) -%}
+{%- for tool in tool_list -%}
+<tool>{{ tool.function | tojson(ensure_ascii=False) }}</tool>
+{% endfor -%}
+{%- endmacro -%}
+{%- 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 -%}
+{#- System Message Construction ============================================ -#}
+{%- macro build_system_message(system_message) -%}
+    {%- if system_message and system_message.content -%}
+        {{- visible_text(system_message.content) }}
+    {%- else -%}
+        {%- if model_identity is not defined -%}
+            {%- set model_identity = "You are a helpful assistant. Your name is MiniMax-M2.7 and is built by MiniMax." -%}
+        {%- endif -%}
+        {{- model_identity }}
+    {%- endif -%}
+    
+    {#- Handle current_date -#}
+    {%- if system_message and system_message.current_date -%}
+        {{- '\n' ~ 'Current date: ' + system_message.current_date }}
+    {%- endif -%}
+    {#- Handle current_location -#}
+    {%- if system_message and system_message.current_location -%}
+        {{- '\n' ~ 'Current location: ' + system_message.current_location }}
+    {%- endif -%}
+{%- endmacro -%}
+{#- Main Template Logic ================================================= -#}
+{#- Extract system message (only first message if it's system) -#}
+{%- set system_message = none -%}
+{%- set conversation_messages = messages -%}
+{%- if messages and messages[0].role == "system" -%}
+    {%- set system_message = messages[0] -%}
+    {%- set conversation_messages = messages[1:] -%}
+{%- endif -%}
+{#- Get the last user message turn, for interleved thinking -#}
+{%- set ns = namespace(last_user_index=-1) %}
+{% for m in conversation_messages %}
+    {%- if m.role == 'user' %}
+        {% set ns.last_user_index = loop.index0 -%}
+    {%- endif %}
+{%- endfor %}
+{#- Render system message -#}
+{{- ']~!b[' ~ ']~b]system' ~ '\n' }}
+{{- build_system_message(system_message) }}
+{#- Render tools if available -#}
+{%- if tools -%}
+    {{- '\n\n' ~ '# Tools' ~ '\n' ~ 'You may call one or more tools to assist with the user query.\nHere are the tools available in JSONSchema format:' ~ '\n' }}
+    {{- '\n' ~ '<tools>' ~ '\n' }}
+    {{- render_tool_namespace("functions", tools) }}
+    {{- '</tools>' ~ '\n\n' }}
+{{- 'When making tool calls, use XML format to invoke tools and pass parameters:' ~ '\n' }}
+{{- '\n' ~ toolcall_begin_token }}
+<invoke name="tool-name-1">
+<parameter name="param-key-1">param-value-1</parameter>
+<parameter name="param-key-2">param-value-2</parameter>
+...
+</invoke>
+{{- '\n' ~ toolcall_end_token }}
+{%- endif -%}
+{{- '[e~[\n' }}
+
+{#- Render messages -#}
+{%- set last_tool_call = namespace(name=none) -%}
+{%- for message in conversation_messages -%}
+    {%- if message.role == 'assistant' -%}
+        {#- Only render reasoning_content if no user message follows -#}
+        {{- ']~b]ai' ~ '\n' }}
+
+        {%- set reasoning_content = '' %}
+        {%- set content = visible_text(message.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].strip('\n').split('<think>')[-1].strip('\n') %}
+                {%- set content = content.split('</think>')[-1].strip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if reasoning_content and loop.index0 > ns.last_user_index -%}
+            {{- '<think>' ~ '\n' ~ reasoning_content ~ '\n' ~ '</think>' ~ '\n\n' }}
+        {%- endif -%}
+        {%- if content -%}
+            {{- content }}
+        {%- endif -%}
+        {%- if message.tool_calls -%}
+            {{- '\n' ~ toolcall_begin_token ~ '\n' }}
+
+            {%- for tool_call in message.tool_calls -%}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<invoke name="' + tool_call.name + '">' }}
+                {% set _args = tool_call.arguments %}
+                {%- for k, v in _args.items() %}
+                {{- '<parameter name="' + k + '">' }}
+                {{- v | tojson(ensure_ascii=False) if v is not string else v }}
+                {{- '</parameter>' }}
+                {% endfor %}
+                {{- '</invoke>' ~ '\n' }}
+            {%- endfor -%}
+            
+            {{- toolcall_end_token}}
+            {%- set last_tool_call.name = message.tool_calls[-1].name -%}
+        {%- else -%}
+            {%- set last_tool_call.name = none -%}
+        {%- endif -%}
+        {{- '[e~[' ~ '\n' }}
+        
+    {%- elif message.role == 'tool' -%}
+    {%- if last_tool_call.name is none -%}
+        {{- raise_exception("Message has tool role, but there was no previous assistant message with a tool call!") }}
+    {%- endif -%}
+    {%- if loop.first or (conversation_messages[loop.index0 - 1].role != 'tool') -%}
+        {{- ']~b]tool' }}
+    {%- endif -%}
+    {%- if message.content is string -%}
+        {{- '\n<response>' }}
+        {{- message.content }}
+        {{- '</response>' }}
+    {%- else -%}
+        {%- for tr in message.content -%}
+            {{- '\n<response>' }}
+            {{- tr.output if tr.output is defined else (tr.text if tr.type == 'text' and tr.text is defined else tr) }}
+            {{- '\n</response>' }}
+        {%- endfor -%}
+    {%- endif -%}
+    {%- if loop.last or (conversation_messages[loop.index0 + 1].role != 'tool') -%}
+        {{- '[e~[\n' -}}
+    {%- endif -%}
+        
+    {%- elif message.role == 'user' -%}
+        {{- ']~b]user' ~ '\n' }}
+        {{- visible_text(message.content) }}
+        {{- '[e~[' ~ '\n' }}
+    {%- endif -%}
+{%- endfor -%}
+
+{#- Generation prompt -#}
+{%- if add_generation_prompt -%}
+{{- ']~b]ai' ~ '\n' ~ '<think>' ~ '\n' }}
+{%- endif -%}

config.json

@@ -0,0 +1,722 @@
+{
+  "architectures": [
+    "MiniMaxM2ForCausalLM"
+  ],
+  "attn_type_list": [
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1,
+    1
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_minimax_m2.MiniMaxM2Config",
+    "AutoModelForCausalLM": "modeling_minimax_m2.MiniMaxM2ForCausalLM"
+  },
+  "dtype": "bfloat16",
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 3072,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 204800,
+  "model_type": "minimax_m2",
+  "mtp_transformer_layers": 1,
+  "num_attention_heads": 48,
+  "num_experts_per_tok": 8,
+  "num_hidden_layers": 62,
+  "num_key_value_heads": 8,
+  "num_local_experts": 176,
+  "num_mtp_modules": 3,
+  "qk_norm_type": "per_layer",
+  "quantization_config": {
+    "quant_method": "compressed-tensors",
+    "format": "mixed-precision",
+    "config_groups": {
+      "group_0": {
+        "format": "float-quantized",
+        "weights": {
+          "num_bits": 8,
+          "type": "float",
+          "strategy": "block",
+          "block_structure": [
+            128,
+            128
+          ],
+          "symmetric": true,
+          "dynamic": false,
+          "observer": "memoryless_minmax"
+        },
+        "input_activations": {
+          "num_bits": 8,
+          "type": "float",
+          "strategy": "tensor",
+          "symmetric": true,
+          "dynamic": true,
+          "observer": "memoryless_minmax"
+        },
+        "targets": [
+          "re:^model[.]layers[.]0[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]0[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]0[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]0[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]0[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]1[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]1[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]1[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]1[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]1[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]11[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]11[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]11[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]11[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]14[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]14[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]14[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]14[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]15[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]15[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]15[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]15[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]16[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]16[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]16[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]16[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]17[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]17[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]17[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]17[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]18[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]18[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]18[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]18[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]19[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]19[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]19[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]19[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]20[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]20[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]20[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]20[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]21[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]21[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]21[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]21[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]22[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]22[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]22[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]22[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]23[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]23[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]23[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]23[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]24[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]24[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]24[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]24[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]25[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]25[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]25[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]25[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]26[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]26[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]26[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]26[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]27[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]27[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]27[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]27[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]28[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]28[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]28[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]28[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]29[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]29[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]29[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]29[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]30[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]30[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]30[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]30[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]31[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]31[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]31[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]31[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]32[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]32[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]32[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]32[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]33[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]33[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]33[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]33[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]34[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]34[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]34[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]34[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]35[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]35[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]35[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]35[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]37[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]37[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]37[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]37[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]39[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]39[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]39[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]39[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]42[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]42[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]42[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]42[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]48[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]48[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]48[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]48[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]49[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]49[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]49[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]49[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]0[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]0[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]0[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]1[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]1[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]1[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]5[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]5[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]5[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]7[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]7[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]7[.]mlp[.]experts[.][0-9]+[.]up_proj$"
+        ]
+      },
+      "group_1": {
+        "format": "nvfp4-pack-quantized",
+        "weights": {
+          "num_bits": 4,
+          "type": "float",
+          "strategy": "tensor_group",
+          "group_size": 16,
+          "symmetric": true,
+          "dynamic": false,
+          "scale_dtype": "torch.float8_e4m3fn",
+          "zp_dtype": "torch.float8_e4m3fn",
+          "observer": "memoryless_minmax"
+        },
+        "input_activations": {
+          "num_bits": 4,
+          "type": "float",
+          "strategy": "tensor_group",
+          "group_size": 16,
+          "symmetric": true,
+          "dynamic": "local",
+          "observer": "static_minmax",
+          "scale_dtype": "torch.float8_e4m3fn",
+          "zp_dtype": "torch.float8_e4m3fn"
+        },
+        "targets": [
+          "re:^model[.]layers[.]10[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]10[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]10[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]10[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]10[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]11[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]12[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]12[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]12[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]12[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]12[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]13[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]13[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]13[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]13[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]13[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]14[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]15[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]16[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]17[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]18[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]19[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]2[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]2[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]2[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]2[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]2[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]20[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]21[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]22[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]23[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]24[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]25[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]26[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]27[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]28[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]29[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]3[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]3[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]3[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]3[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]3[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]30[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]31[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]32[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]33[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]34[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]35[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]36[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]36[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]36[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]36[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]36[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]37[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]38[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]38[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]38[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]38[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]38[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]39[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]4[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]4[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]4[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]4[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]4[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]40[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]40[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]40[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]40[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]40[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]41[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]41[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]41[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]41[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]41[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]42[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]43[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]43[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]43[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]43[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]43[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]44[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]44[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]44[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]44[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]44[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]45[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]45[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]45[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]45[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]45[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]46[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]46[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]46[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]46[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]46[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]47[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]47[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]47[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]47[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]47[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]48[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]49[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]5[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]5[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]5[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]5[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]5[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]50[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]50[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]50[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]50[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]50[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]51[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]51[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]51[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]51[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]51[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]52[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]52[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]52[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]52[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]52[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]53[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]53[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]53[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]53[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]53[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]54[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]54[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]54[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]54[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]54[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]55[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]55[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]55[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]55[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]55[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]56[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]56[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]56[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]56[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]56[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]57[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]57[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]57[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]57[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]57[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]58[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]58[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]58[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]58[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]58[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]59[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]59[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]59[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]59[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]59[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]6[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]6[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]6[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]6[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]6[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]60[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]60[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]60[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]60[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]60[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]61[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]61[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]61[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]61[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]61[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]7[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]7[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]7[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]7[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]7[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]8[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]8[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]8[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]8[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]8[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]9[.]self_attn[.]k_proj$",
+          "re:^model[.]layers[.]9[.]self_attn[.]o_proj$",
+          "re:^model[.]layers[.]9[.]self_attn[.]q_proj$",
+          "re:^model[.]layers[.]9[.]self_attn[.]qkv_proj$",
+          "re:^model[.]layers[.]9[.]self_attn[.]v_proj$",
+          "re:^model[.]layers[.]10[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]10[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]10[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]11[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]11[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]11[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]12[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]12[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]12[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]13[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]13[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]13[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]14[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]14[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]14[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]15[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]15[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]15[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]16[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]16[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]16[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]17[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]17[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]17[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]18[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]18[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]18[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]19[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]19[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]19[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]20[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]20[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]20[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]21[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]21[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]21[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]22[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]22[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]22[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]23[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]23[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]23[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]24[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]24[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]24[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]25[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]25[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]25[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]26[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]26[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]26[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]27[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]27[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]27[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]28[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]28[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]28[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]29[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]29[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]29[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]2[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]2[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]2[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]30[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]30[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]30[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]31[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]31[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]31[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]32[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]32[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]32[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]33[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]33[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]33[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]34[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]34[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]34[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]35[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]35[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]35[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]36[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]36[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]36[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]37[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]37[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]37[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]38[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]38[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]38[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]39[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]39[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]39[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]3[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]3[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]3[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]40[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]40[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]40[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]41[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]41[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]41[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]42[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]42[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]42[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]43[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]43[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]43[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]44[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]44[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]44[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]45[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]45[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]45[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]46[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]46[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]46[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]47[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]47[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]47[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]48[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]48[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]48[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]49[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]49[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]49[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]4[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]4[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]4[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]50[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]50[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]50[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]51[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]51[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]51[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]52[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]52[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]52[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]53[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]53[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]53[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]54[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]54[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]54[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]55[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]55[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]55[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]56[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]56[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]56[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]57[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]57[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]57[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]58[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]58[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]58[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]59[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]59[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]59[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]60[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]60[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]60[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]61[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]61[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]61[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]6[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]6[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]6[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]8[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]8[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]8[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.]9[.]mlp[.]experts[.][0-9]+[.]down_proj$",
+          "re:^model[.]layers[.]9[.]mlp[.]experts[.][0-9]+[.]gate_proj$",
+          "re:^model[.]layers[.]9[.]mlp[.]experts[.][0-9]+[.]up_proj$",
+          "re:^model[.]layers[.][0-9]+[.]mlp[.]experts[.][0-9]+[.](gate|up|down)_proj$"
+        ]
+      }
+    },
+    "ignore": [
+      "lm_head",
+      "model.embed_tokens",
+      "model.layers.0.mlp.gate",
+      "model.layers.1.mlp.gate",
+      "model.layers.10.mlp.gate",
+      "model.layers.11.mlp.gate",
+      "model.layers.12.mlp.gate",
+      "model.layers.13.mlp.gate",
+      "model.layers.14.mlp.gate",
+      "model.layers.15.mlp.gate",
+      "model.layers.16.mlp.gate",
+      "model.layers.17.mlp.gate",
+      "model.layers.18.mlp.gate",
+      "model.layers.19.mlp.gate",
+      "model.layers.2.mlp.gate",
+      "model.layers.20.mlp.gate",
+      "model.layers.21.mlp.gate",
+      "model.layers.22.mlp.gate",
+      "model.layers.23.mlp.gate",
+      "model.layers.24.mlp.gate",
+      "model.layers.25.mlp.gate",
+      "model.layers.26.mlp.gate",
+      "model.layers.27.mlp.gate",
+      "model.layers.28.mlp.gate",
+      "model.layers.29.mlp.gate",
+      "model.layers.3.mlp.gate",
+      "model.layers.30.mlp.gate",
+      "model.layers.31.mlp.gate",
+      "model.layers.32.mlp.gate",
+      "model.layers.33.mlp.gate",
+      "model.layers.34.mlp.gate",
+      "model.layers.35.mlp.gate",
+      "model.layers.36.mlp.gate",
+      "model.layers.37.mlp.gate",
+      "model.layers.38.mlp.gate",
+      "model.layers.39.mlp.gate",
+      "model.layers.4.mlp.gate",
+      "model.layers.40.mlp.gate",
+      "model.layers.41.mlp.gate",
+      "model.layers.42.mlp.gate",
+      "model.layers.43.mlp.gate",
+      "model.layers.44.mlp.gate",
+      "model.layers.45.mlp.gate",
+      "model.layers.46.mlp.gate",
+      "model.layers.47.mlp.gate",
+      "model.layers.48.mlp.gate",
+      "model.layers.49.mlp.gate",
+      "model.layers.5.mlp.gate",
+      "model.layers.50.mlp.gate",
+      "model.layers.51.mlp.gate",
+      "model.layers.52.mlp.gate",
+      "model.layers.53.mlp.gate",
+      "model.layers.54.mlp.gate",
+      "model.layers.55.mlp.gate",
+      "model.layers.56.mlp.gate",
+      "model.layers.57.mlp.gate",
+      "model.layers.58.mlp.gate",
+      "model.layers.59.mlp.gate",
+      "model.layers.6.mlp.gate",
+      "model.layers.60.mlp.gate",
+      "model.layers.61.mlp.gate",
+      "model.layers.7.mlp.gate",
+      "model.layers.8.mlp.gate",
+      "model.layers.9.mlp.gate"
+    ],
+    "quantization_status": "compressed"
+  },
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 5000000,
+  "rotary_dim": 64,
+  "scoring_func": "sigmoid",
+  "shared_intermediate_size": 0,
+  "tie_word_embeddings": false,
+  "transformers_version": "4.46.1",
+  "use_cache": true,
+  "use_mtp": true,
+  "use_qk_norm": true,
+  "use_routing_bias": true,
+  "vocab_size": 200064
+}

configuration_minimax_m2.py

@@ -0,0 +1,200 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from src/transformers/models/minimax_m2/modular_minimax_m2.py.
+#               Do NOT edit this file manually as any edits will be overwritten by the generation of
+#             the file from the modular. If any change should be done, please apply the change to the
+#                          modular_minimax_m2.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# coding=utf-8
+# Copyright 2025 the HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+from transformers.configuration_utils import PretrainedConfig
+
+
+class MiniMaxM2Config(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`MiniMaxM2Model`]. It is used to instantiate an
+    MiniMaxM2 model according to the specified arguments, defining the model architecture. Instantiating a configuration
+    with the defaults will yield a similar configuration to that of the MiniMaxM2-7B-v0.1 or MiniMaxM2-7B-Instruct-v0.1.
+
+    [minimax_m2ai/MiniMaxM2-8x7B](https://huggingface.co/minimax_m2ai/MiniMaxM2-8x7B)
+    [minimax_m2ai/MiniMaxM2-7B-Instruct-v0.1](https://huggingface.co/minimax_m2ai/MiniMaxM2-7B-Instruct-v0.1)
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 32000):
+            Vocabulary size of the MiniMaxM2 model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`MiniMaxM2Model`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 14336):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        num_key_value_heads (`int`, *optional*, defaults to 8):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details, check out [this
+            paper](https://huggingface.co/papers/2305.13245). If it is not specified, will default to `8`.
+        head_dim (`int`, *optional*, defaults to `hidden_size // num_attention_heads`):
+            The attention head dimension.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to `4096*32`):
+            The maximum sequence length that this model might ever be used with. MiniMaxM2's sliding window attention
+            allows sequence of up to 4096*32 tokens.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-05):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        pad_token_id (`int`, *optional*):
+            The id of the padding token.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            The id of the "beginning-of-sequence" token.
+        eos_token_id (`int`, *optional*, defaults to 2):
+            The id of the "end-of-sequence" token.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether the model's input and output word embeddings should be tied.
+        rope_theta (`float`, *optional*, defaults to 1000000.0):
+            The base period of the RoPE embeddings.
+        sliding_window (`int`, *optional*):
+            Sliding window attention window size. If not specified, will default to `4096`.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        num_experts_per_tok (`int`, *optional*, defaults to 2):
+            The number of experts to route per-token, can be also interpreted as the `top-k` routing
+            parameter
+        num_local_experts (`int`, *optional*, defaults to 8):
+            Number of experts per Sparse MLP layer.
+        output_router_logits (`bool`, *optional*, defaults to `False`):
+            Whether or not the router logits should be returned by the model. Enabling this will also
+            allow the model to output the auxiliary loss. See [here]() for more details
+        router_aux_loss_coef (`float`, *optional*, defaults to 0.001):
+            The aux loss factor for the total loss.
+        router_jitter_noise (`float`, *optional*, defaults to 0.0):
+            Amount of noise to add to the router.
+
+    ```python
+    >>> from transformers import MiniMaxM2Model, MiniMaxM2Config
+
+    >>> # Initializing a MiniMaxM2 7B style configuration
+    >>> configuration = MiniMaxM2Config()
+
+    >>> # Initializing a model from the MiniMaxM2 7B style configuration
+    >>> model = MiniMaxM2Model(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "minimax_m2"
+    keys_to_ignore_at_inference = ["past_key_values"]
+    base_model_tp_plan = {
+        "layers.*.self_attn.q_proj": "colwise",
+        "layers.*.self_attn.k_proj": "colwise",
+        "layers.*.self_attn.v_proj": "colwise",
+        "layers.*.self_attn.o_proj": "rowwise",
+        "layers.*.block_sparse_moe.gate": "colwise_rep",  # we need to replicate here to correctly route experts
+        "layers.*.block_sparse_moe.experts.*.w1": "colwise",
+        "layers.*.block_sparse_moe.experts.*.w2": "rowwise",
+        "layers.*.block_sparse_moe.experts.*.w3": "colwise",
+    }
+    base_model_pp_plan = {
+        "embed_tokens": (["input_ids"], ["inputs_embeds"]),
+        "layers": (["hidden_states", "attention_mask"], ["hidden_states"]),
+        "norm": (["hidden_states"], ["hidden_states"]),
+    }
+
+    def __init__(
+        self,
+        vocab_size=32000,
+        hidden_size=4096,
+        intermediate_size=14336,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=8,
+        head_dim=None,
+        hidden_act="silu",
+        max_position_embeddings=4096 * 32,
+        initializer_range=0.02,
+        rms_norm_eps=1e-5,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=1,
+        eos_token_id=2,
+        tie_word_embeddings=False,
+        rope_theta=1e6,
+        sliding_window=None,
+        attention_dropout=0.0,
+        num_experts_per_tok=2,
+        num_local_experts=8,
+        output_router_logits=False,
+        router_aux_loss_coef=0.001,
+        router_jitter_noise=0.0,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.sliding_window = sliding_window
+
+        # for backward compatibility
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+
+        self.num_key_value_heads = num_key_value_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.attention_dropout = attention_dropout
+        self.head_dim = head_dim
+
+        self.num_experts_per_tok = num_experts_per_tok
+        self.num_local_experts = num_local_experts
+        self.output_router_logits = output_router_logits
+        self.router_aux_loss_coef = router_aux_loss_coef
+        self.router_jitter_noise = router_jitter_noise
+
+        self.use_qk_norm = kwargs.pop("use_qk_norm", False)
+        self.rotary_dim = kwargs.pop("rotary_dim", self.head_dim)
+        self.partial_rotary_factor = kwargs.pop("partial_rotary_factor", 1)
+        if self.head_dim is not None:
+            self.partial_rotary_factor = self.rotary_dim / self.head_dim
+
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+
+
+__all__ = ["MiniMaxM2Config"]

generation_config.json

@@ -0,0 +1,9 @@
+{
+  "bos_token_id": 200019,
+  "do_sample": true,
+  "eos_token_id": 200020,
+  "temperature": 1.0,
+  "top_p": 0.95,
+  "top_k": 40,
+  "transformers_version": "4.46.1"
+}

mixed_native_manifest.json

@@ -0,0 +1,23 @@
+{
+  "source_model": "/hfcache/hub/models--MiniMaxAI--MiniMax-M2.7/snapshots/d494266a4affc0d2995ba1fa35c8481cbd84294b",
+  "source_recipe": "/work/artifacts/layer_config_prune_t320.json",
+  "format_histogram": {
+    "head_passthrough/BF16": 3,
+    "linear/FP8_SOURCE": 2204,
+    "linear/BF16_router_shrunk": 62,
+    "linear/PRUNED": 14880,
+    "layer_passthrough/BF16": 248,
+    "layer_buffer_shrunk/BF16": 62,
+    "linear/NVFP4": 30780
+  },
+  "n_assignment_entries": 32984,
+  "ignore": [
+    "lm_head"
+  ],
+  "prune": {
+    "n_layers_pruned": 62,
+    "n_experts_orig_total": 15872,
+    "n_experts_kept_total": 10912,
+    "manifest_file": "prune_manifest.json"
+  }
+}

model-00001-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1752e5fc0bbb9039e2b68f95e3cdf6635b09436c0c992b34b2add1212f91903
+size 5368595808

model-00002-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4627caae229f91a784404cea5f208207e2716589fb43a6f73e110ce18d44c358
+size 5367391056

model-00003-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10cbb451d5436387ac69d5d632c9ec6cf0481014850793e0ee5ab477e3608999
+size 5365284976

model-00004-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de3cfc91b0f8a18e306dc7d19b7c43d1d49e4dd2cd41fe15b03517de33c7c8e9
+size 5367878440

model-00005-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa18b2c61b863d5e10cd442f6ecd56fff257ae1bffad34c4c03fa1c0e0455068
+size 5363845816

model-00006-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47dfed04bc341c05542cbe0e6d37c23379499eabf30dee8592e9874c20c2f1b2
+size 5368863864

model-00007-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:376ac49cf00ab81688edbe13dd47826c5787ecbca8da54843c35b2732a6e8204
+size 5369471952

model-00008-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:318d820c0b276068f90e3f806fe59e3c3172f4e8bff6f77b5118515eeb674fad
+size 5369471344

model-00009-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6761a778d45c623075ae9931007ed4d33a6fc05aeec11cbf642d69e700f8a2f3
+size 5369738888

model-00010-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20f89480389d93def5c4a23560da6c40db0f165ea9c8beeaa6a1f7b20a720d83
+size 5369472392

model-00011-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca3c2387d19852ce9c11f42b4cbcfe511763ec7892a1638e8068d736224ea094
+size 5369471808

model-00012-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51d7dc536f60978039a02fe9f4e997278c2ca06554fe1caf455727bffc6ec3ac
+size 5368676824

model-00013-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7300c93ab277c426fd42a8b7f81a9b110be12314da0b7ee6285a01a2875a0827
+size 5368942944

model-00014-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab54ddbd2733cf62b71cd5aa49bf69f73a21fce2a5ee5490c851b4702e9e21fa
+size 5367882824

model-00015-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d915392faa3e3b509c626d6c7329a9cc2734edac2e9a3cf145e4b86c2a729410
+size 5368677728

model-00016-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28e767c0959449451639165964420651091901f2feba9be658ab8da2fe2772ef
+size 5367882288

model-00017-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d555a575b7bd3abf44d8067b32b8020103afdead0a7c8c2f5b1c1cdcf9cdb7a
+size 5368546424

model-00018-of-00018.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c6f4cf130124a32a56e9307cb13f0f831fe8b49d8cc6c2d833d3393f8c72433
+size 4417483752

model.safetensors.index.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:384eaf9ab722697a63b258c37aeb43c1f7ae3f4bb640fd4c2ad241f2e1e562a5
+size 13605125

modeling_minimax_m2.py

@@ -0,0 +1,706 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from src/transformers/models/minimax_m2/modular_minimax_m2.py.
+#               Do NOT edit this file manually as any edits will be overwritten by the generation of
+#             the file from the modular. If any change should be done, please apply the change to the
+#                          modular_minimax_m2.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# coding=utf-8
+# Copyright 2025 the HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+from collections.abc import Callable
+from typing import Optional, Union, Unpack
+
+import torch
+from torch import nn
+
+from transformers.activations import ACT2FN
+from transformers.cache_utils import Cache, DynamicCache
+from transformers.generation import GenerationMixin
+from transformers.integrations import use_kernel_forward_from_hub
+from transformers.masking_utils import create_causal_mask, create_sliding_window_causal_mask
+from transformers.modeling_flash_attention_utils import FlashAttentionKwargs
+from transformers.modeling_layers import (
+    GenericForQuestionAnswering,
+    GenericForSequenceClassification,
+    GenericForTokenClassification,
+    GradientCheckpointingLayer,
+)
+from transformers.modeling_outputs import MoeCausalLMOutputWithPast, MoeModelOutputWithPast
+from transformers.modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
+from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
+from transformers.utils import TransformersKwargs, auto_docstring, can_return_tuple
+from transformers.utils.deprecation import deprecate_kwarg
+from transformers.utils.generic import OutputRecorder, check_model_inputs
+from .configuration_minimax_m2 import MiniMaxM2Config
+
+
+class MiniMaxM2MLP(nn.Module):
+    def __init__(self, config: MiniMaxM2Config):
+        super().__init__()
+        self.ffn_dim = config.intermediate_size
+        self.hidden_dim = config.hidden_size
+
+        self.w1 = nn.Linear(self.hidden_dim, self.ffn_dim, bias=False)
+        self.w2 = nn.Linear(self.ffn_dim, self.hidden_dim, bias=False)
+        self.w3 = nn.Linear(self.hidden_dim, self.ffn_dim, bias=False)
+
+        self.act_fn = ACT2FN[config.hidden_act]
+
+    def forward(self, hidden_states):
+        current_hidden_states = self.act_fn(self.w1(hidden_states)) * self.w3(hidden_states)
+        current_hidden_states = self.w2(current_hidden_states)
+        return current_hidden_states
+
+
+class MiniMaxM2Experts(nn.ModuleList):
+    """
+    ModuleList of experts.
+    """
+
+    def __init__(self, config: MiniMaxM2Config):
+        super().__init__()
+        self.top_k = config.num_experts_per_tok
+        self.num_experts = config.num_local_experts
+        for _ in range(self.num_experts):
+            self.append(MiniMaxM2MLP(config))
+
+    def forward(
+        self, hidden_states: torch.Tensor, top_k_index: torch.Tensor, top_k_weights: torch.Tensor
+    ) -> torch.Tensor:
+        """
+        Args:
+            hidden_states: (batch_size * sequence_length, hidden_dim)
+            selected_experts: (batch_size * sequence_length, top_k)
+            routing_weights: (batch_size * sequence_length, top_k)
+        Returns:
+            (batch_size * sequence_length, hidden_dim)
+        """
+        final_hidden_states = torch.zeros_like(hidden_states)
+        expert_mask = torch.nn.functional.one_hot(top_k_index, num_classes=self.num_experts).permute(2, 1, 0)
+
+        expert_hit = torch.greater(expert_mask.sum(dim=(-1, -2)), 0).nonzero()
+        for expert_idx in expert_hit:
+            idx, top_x = torch.where(expert_mask[expert_idx].squeeze(0))
+            current_state = hidden_states[None, top_x].reshape(-1, hidden_states.shape[-1])
+            current_hidden_states = self[expert_idx](current_state) * top_k_weights[top_x, idx, None]
+            final_hidden_states.index_add_(0, top_x, current_hidden_states.to(hidden_states.dtype))
+        return final_hidden_states
+
+
+class MiniMaxM2SparseMoeBlock(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.top_k = config.num_experts_per_tok
+        self.jitter_noise = config.router_jitter_noise
+        self.gate = nn.Linear(config.hidden_size, config.num_local_experts, bias=False)
+        self.experts = MiniMaxM2Experts(config)
+        self.register_buffer("e_score_correction_bias", torch.zeros(config.num_local_experts))
+
+    def route_tokens_to_experts(self, router_logits):
+        routing_weights = torch.nn.functional.sigmoid(router_logits.float())
+        scores_for_choice = routing_weights + self.e_score_correction_bias
+        _, top_k_index = torch.topk(scores_for_choice, self.top_k, dim=-1, sorted=False)
+        top_k_weights = routing_weights.gather(1, top_k_index)
+        top_k_weights /= top_k_weights.sum(dim=-1, keepdim=True)
+        return top_k_index, top_k_weights.to(router_logits.dtype)
+
+    def forward(self, hidden_states: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
+        batch_size, sequence_length, hidden_dim = hidden_states.shape
+        if self.training and self.jitter_noise > 0:
+            hidden_states *= torch.empty_like(hidden_states).uniform_(1.0 - self.jitter_noise, 1.0 + self.jitter_noise)
+        hidden_states = hidden_states.view(-1, hidden_states.shape[-1])
+        router_logits = self.gate(hidden_states)
+        top_k_index, top_k_weights = self.route_tokens_to_experts(router_logits)
+        hidden_states = self.experts(hidden_states, top_k_index, top_k_weights.to(hidden_states.dtype))
+        hidden_states = hidden_states.reshape(batch_size, sequence_length, hidden_dim)
+        return hidden_states, router_logits
+
+
+@use_kernel_forward_from_hub("RMSNorm")
+class MiniMaxM2RMSNorm(nn.Module):
+    def __init__(self, hidden_size, eps=1e-6):
+        """
+        MiniMaxM2RMSNorm is equivalent to T5LayerNorm
+        """
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        input_dtype = hidden_states.dtype
+        hidden_states = hidden_states.to(torch.float32)
+        variance = hidden_states.pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+        return self.weight * hidden_states.to(input_dtype)
+
+    def extra_repr(self):
+        return f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}"
+
+
+def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
+    """
+    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
+    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
+    """
+    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
+    if n_rep == 1:
+        return hidden_states
+    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
+    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
+
+
+def eager_attention_forward(
+    module: nn.Module,
+    query: torch.Tensor,
+    key: torch.Tensor,
+    value: torch.Tensor,
+    attention_mask: Optional[torch.Tensor],
+    scaling: float,
+    dropout: float = 0.0,
+    **kwargs: Unpack[TransformersKwargs],
+):
+    key_states = repeat_kv(key, module.num_key_value_groups)
+    value_states = repeat_kv(value, module.num_key_value_groups)
+
+    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
+    if attention_mask is not None:
+        causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
+        attn_weights = attn_weights + causal_mask
+
+    attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
+    attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)
+    attn_output = torch.matmul(attn_weights, value_states)
+    attn_output = attn_output.transpose(1, 2).contiguous()
+
+    return attn_output, attn_weights
+
+
+def rotate_half(x):
+    """Rotates half the hidden dims of the input."""
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    return torch.cat((-x2, x1), dim=-1)
+
+
+def apply_rotary_pos_emb(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
+    """Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q (`torch.Tensor`): The query tensor.
+        k (`torch.Tensor`): The key tensor.
+        cos (`torch.Tensor`): The cosine part of the rotary embedding.
+        sin (`torch.Tensor`): The sine part of the rotary embedding.
+        position_ids (`torch.Tensor`, *optional*):
+            Deprecated and unused.
+        unsqueeze_dim (`int`, *optional*, defaults to 1):
+            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
+            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
+            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
+            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
+            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
+            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
+    Returns:
+        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
+    """
+    cos = cos.unsqueeze(unsqueeze_dim)
+    sin = sin.unsqueeze(unsqueeze_dim)
+
+    # Keep half or full tensor for later concatenation
+    rotary_dim = cos.shape[-1]
+    q_rot, q_pass = q[..., :rotary_dim], q[..., rotary_dim:]
+    k_rot, k_pass = k[..., :rotary_dim], k[..., rotary_dim:]
+
+    # Apply rotary embeddings on the first half or full tensor
+    q_embed = (q_rot * cos) + (rotate_half(q_rot) * sin)
+    k_embed = (k_rot * cos) + (rotate_half(k_rot) * sin)
+
+    # Concatenate back to full shape
+    q_embed = torch.cat([q_embed, q_pass], dim=-1)
+    k_embed = torch.cat([k_embed, k_pass], dim=-1)
+    return q_embed, k_embed
+
+
+class MiniMaxM2Attention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: MiniMaxM2Config, layer_idx: int):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        self.head_dim = getattr(config, "head_dim", None) or config.hidden_size // config.num_attention_heads
+        self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
+        self.scaling = self.head_dim**-0.5
+        self.attention_dropout = config.attention_dropout
+        self.is_causal = True
+        self.q_proj = nn.Linear(config.hidden_size, config.num_attention_heads * self.head_dim, bias=False)
+        self.k_proj = nn.Linear(config.hidden_size, config.num_key_value_heads * self.head_dim, bias=False)
+        self.v_proj = nn.Linear(config.hidden_size, config.num_key_value_heads * self.head_dim, bias=False)
+        self.o_proj = nn.Linear(config.num_attention_heads * self.head_dim, config.hidden_size, bias=False)
+
+        self.use_qk_norm = config.use_qk_norm
+        if self.use_qk_norm:
+            self.q_norm = MiniMaxM2RMSNorm(self.head_dim * config.num_attention_heads, eps=config.rms_norm_eps)
+            self.k_norm = MiniMaxM2RMSNorm(self.head_dim * config.num_key_value_heads, eps=config.rms_norm_eps)
+
+    @deprecate_kwarg("past_key_value", new_name="past_key_values", version="4.58")
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        position_embeddings: tuple[torch.Tensor, torch.Tensor],
+        attention_mask: Optional[torch.Tensor],
+        past_key_values: Optional[Cache] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs: Unpack[FlashAttentionKwargs],
+    ) -> tuple[torch.Tensor, Optional[torch.Tensor]]:
+        input_shape = hidden_states.shape[:-1]
+        hidden_shape = (*input_shape, -1, self.head_dim)
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        if self.use_qk_norm:  # main diff from Llama
+            query_states = self.q_norm(query_states)
+            key_states = self.k_norm(key_states)
+
+        key_states = key_states.view(hidden_shape)
+        query_states = query_states.view(hidden_shape)
+        value_states = value_states.view(hidden_shape)
+
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        cos, sin = position_embeddings
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
+
+        if past_key_values is not None:
+            # sin and cos are specific to RoPE models; position_ids needed for the static cache
+            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
+            key_states, value_states = past_key_values.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        attention_interface: Callable = eager_attention_forward
+        if self.config._attn_implementation != "eager":
+            attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
+
+        attn_output, attn_weights = attention_interface(
+            self,
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            dropout=0.0 if not self.training else self.attention_dropout,
+            scaling=self.scaling,
+            **kwargs,
+        )
+
+        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
+        attn_output = self.o_proj(attn_output)
+        return attn_output, attn_weights
+
+
+class MiniMaxM2DecoderLayer(GradientCheckpointingLayer):
+    def __init__(self, config: MiniMaxM2Config, layer_idx: int):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+
+        self.self_attn = MiniMaxM2Attention(config, layer_idx)
+
+        self.block_sparse_moe = MiniMaxM2SparseMoeBlock(config)
+        self.input_layernorm = MiniMaxM2RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.post_attention_layernorm = MiniMaxM2RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    @deprecate_kwarg("past_key_value", new_name="past_key_values", version="4.58")
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        position_embeddings: tuple[torch.Tensor, torch.Tensor],
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> torch.FloatTensor:
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states)
+
+        # Self Attention
+        hidden_states, _ = self.self_attn(
+            hidden_states=hidden_states,
+            position_embeddings=position_embeddings,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            cache_position=cache_position,
+            **kwargs,
+        )
+        hidden_states = residual + hidden_states
+
+        # Fully Connected
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states, _ = self.block_sparse_moe(hidden_states)
+        hidden_states = residual + hidden_states
+
+        return hidden_states
+
+
+class MiniMaxM2RotaryEmbedding(nn.Module):
+    inv_freq: torch.Tensor  # fix linting for `register_buffer`
+
+    def __init__(self, config: MiniMaxM2Config, device=None):
+        super().__init__()
+        # BC: "rope_type" was originally "type"
+        if hasattr(config, "rope_scaling") and isinstance(config.rope_scaling, dict):
+            self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
+        else:
+            self.rope_type = "default"
+        self.max_seq_len_cached = config.max_position_embeddings
+        self.original_max_seq_len = config.max_position_embeddings
+
+        self.config = config
+        self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
+
+        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
+        self.register_buffer("inv_freq", inv_freq, persistent=False)
+        self.original_inv_freq = self.inv_freq
+
+    @torch.no_grad()
+    @dynamic_rope_update  # power user: used with advanced RoPE types (e.g. dynamic rope)
+    def forward(self, x, position_ids):
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
+        position_ids_expanded = position_ids[:, None, :].float()
+
+        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
+        with torch.autocast(device_type=device_type, enabled=False):  # Force float32
+            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
+            emb = torch.cat((freqs, freqs), dim=-1)
+            cos = emb.cos() * self.attention_scaling
+            sin = emb.sin() * self.attention_scaling
+
+        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)
+
+
+@auto_docstring
+class MiniMaxM2PreTrainedModel(PreTrainedModel):
+    config: MiniMaxM2Config
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["MiniMaxM2DecoderLayer"]
+    _skip_keys_device_placement = ["past_key_values"]
+    _supports_flash_attn = True
+    _supports_sdpa = True
+    _supports_flex_attn = True
+    _can_compile_fullgraph = False  # MoE models don't work with torch.compile (`torch.where(condition)` not supported)
+    _supports_attention_backend = True
+    _can_record_outputs = {
+        "router_logits": OutputRecorder(MiniMaxM2SparseMoeBlock, index=1),
+        "hidden_states": MiniMaxM2DecoderLayer,
+        "attentions": MiniMaxM2Attention,
+    }
+
+
+@auto_docstring
+class MiniMaxM2Model(MiniMaxM2PreTrainedModel):
+    def __init__(self, config: MiniMaxM2Config):
+        super().__init__(config)
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)
+        self.layers = nn.ModuleList(
+            [MiniMaxM2DecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
+        )
+        self.norm = MiniMaxM2RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.rotary_emb = MiniMaxM2RotaryEmbedding(config=config)
+        self.gradient_checkpointing = False
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @check_model_inputs
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> MoeModelOutputWithPast:
+        if (input_ids is None) ^ (inputs_embeds is not None):
+            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
+
+        if use_cache and past_key_values is None:
+            past_key_values = DynamicCache(config=self.config)
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+
+        if cache_position is None:
+            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
+            cache_position = torch.arange(
+                past_seen_tokens, past_seen_tokens + inputs_embeds.shape[1], device=inputs_embeds.device
+            )
+        if position_ids is None:
+            position_ids = cache_position.unsqueeze(0)
+
+        mask_function = create_causal_mask if self.config.sliding_window is None else create_sliding_window_causal_mask
+        causal_mask = mask_function(
+            config=self.config,
+            input_embeds=inputs_embeds,
+            attention_mask=attention_mask,
+            cache_position=cache_position,
+            past_key_values=past_key_values,
+            position_ids=position_ids,
+        )
+
+        hidden_states = inputs_embeds
+
+        # create position embeddings to be shared across the decoder layers
+        position_embeddings = self.rotary_emb(hidden_states, position_ids)
+
+        for decoder_layer in self.layers[: self.config.num_hidden_layers]:
+            hidden_states = decoder_layer(
+                hidden_states,
+                position_embeddings=position_embeddings,
+                attention_mask=causal_mask,
+                position_ids=position_ids,
+                past_key_values=past_key_values,
+                use_cache=use_cache,
+                cache_position=cache_position,
+                **kwargs,
+            )
+
+        hidden_states = self.norm(hidden_states)
+
+        return MoeModelOutputWithPast(  # only diff with Mistral is the output type, we need MoE
+            last_hidden_state=hidden_states,
+            past_key_values=past_key_values,
+        )
+
+
+def load_balancing_loss_func(
+    gate_logits: Union[torch.Tensor, tuple[torch.Tensor], None],
+    num_experts: Optional[int] = None,
+    top_k=2,
+    attention_mask: Optional[torch.Tensor] = None,
+) -> Union[torch.Tensor, int]:
+    r"""
+    Computes auxiliary load balancing loss as in Switch Transformer - implemented in Pytorch.
+
+    See Switch Transformer (https://huggingface.co/papers/2101.03961) for more details. This function implements the loss
+    function presented in equations (4) - (6) of the paper. It aims at penalizing cases where the routing between
+    experts is too unbalanced.
+
+    Args:
+        gate_logits:
+            Logits from the `gate`, should be a tuple of model.config.num_hidden_layers tensors of
+            shape [batch_size X sequence_length, num_experts].
+        num_experts:
+            Number of experts
+        top_k:
+            The number of experts to route per-token, can be also interpreted as the `top-k` routing
+            parameter.
+        attention_mask (`torch.Tensor`, *optional*):
+            The attention_mask used in forward function
+            shape [batch_size X sequence_length] if not None.
+
+    Returns:
+        The auxiliary loss.
+    """
+    if gate_logits is None or not isinstance(gate_logits, tuple):
+        return 0
+
+    if isinstance(gate_logits, tuple):
+        compute_device = gate_logits[0].device
+        concatenated_gate_logits = torch.cat([layer_gate.to(compute_device) for layer_gate in gate_logits], dim=0)
+
+    routing_weights = torch.nn.functional.softmax(concatenated_gate_logits, dim=-1)
+
+    _, selected_experts = torch.topk(routing_weights, top_k, dim=-1)
+
+    expert_mask = torch.nn.functional.one_hot(selected_experts, num_experts)
+
+    if attention_mask is None:
+        # Compute the percentage of tokens routed to each experts
+        tokens_per_expert = torch.mean(expert_mask.float(), dim=0)
+
+        # Compute the average probability of routing to these experts
+        router_prob_per_expert = torch.mean(routing_weights, dim=0)
+    else:
+        batch_size, sequence_length = attention_mask.shape
+        num_hidden_layers = concatenated_gate_logits.shape[0] // (batch_size * sequence_length)
+
+        # Compute the mask that masks all padding tokens as 0 with the same shape of expert_mask
+        expert_attention_mask = (
+            attention_mask[None, :, :, None, None]
+            .expand((num_hidden_layers, batch_size, sequence_length, top_k, num_experts))
+            .reshape(-1, top_k, num_experts)
+            .to(compute_device)
+        )
+
+        # Compute the percentage of tokens routed to each experts
+        tokens_per_expert = torch.sum(expert_mask.float() * expert_attention_mask, dim=0) / torch.sum(
+            expert_attention_mask, dim=0
+        )
+
+        # Compute the mask that masks all padding tokens as 0 with the same shape of tokens_per_expert
+        router_per_expert_attention_mask = (
+            attention_mask[None, :, :, None]
+            .expand((num_hidden_layers, batch_size, sequence_length, num_experts))
+            .reshape(-1, num_experts)
+            .to(compute_device)
+        )
+
+        # Compute the average probability of routing to these experts
+        router_prob_per_expert = torch.sum(routing_weights * router_per_expert_attention_mask, dim=0) / torch.sum(
+            router_per_expert_attention_mask, dim=0
+        )
+
+    overall_loss = torch.sum(tokens_per_expert * router_prob_per_expert.unsqueeze(0))
+    return overall_loss * num_experts
+
+
+@auto_docstring
+class MiniMaxM2ForCausalLM(MiniMaxM2PreTrainedModel, GenerationMixin):
+    _tied_weights_keys = ["lm_head.weight"]
+    _tp_plan = {"lm_head": "colwise_rep"}
+    _pp_plan = {"lm_head": (["hidden_states"], ["logits"])}
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.model = MiniMaxM2Model(config)
+        self.vocab_size = config.vocab_size
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+        self.router_aux_loss_coef = config.router_aux_loss_coef
+        self.num_experts = config.num_local_experts
+        self.num_experts_per_tok = config.num_experts_per_tok
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @can_return_tuple
+    @auto_docstring
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Cache] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_router_logits: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        logits_to_keep: Union[int, torch.Tensor] = 0,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> MoeCausalLMOutputWithPast:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Example:
+
+        ```python
+        >>> from transformers import AutoTokenizer, MiniMaxM2ForCausalLM
+
+        >>> model = MiniMaxM2ForCausalLM.from_pretrained("mistralai/MiniMaxM2-8x7B-v0.1")
+        >>> tokenizer = AutoTokenizer.from_pretrained("mistralai/MiniMaxM2-8x7B-v0.1")
+
+        >>> prompt = "Hey, are you conscious? Can you talk to me?"
+        >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+        >>> # Generate
+        >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+        >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+        ```"""
+
+        output_router_logits = (
+            output_router_logits if output_router_logits is not None else self.config.output_router_logits
+        )
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs: MoeModelOutputWithPast = self.model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_router_logits=output_router_logits,
+            cache_position=cache_position,
+            **kwargs,
+        )
+
+        hidden_states = outputs.last_hidden_state
+        # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
+        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
+        logits = self.lm_head(hidden_states[:, slice_indices, :])
+
+        loss = None
+        if labels is not None:
+            loss = self.loss_function(logits, labels, self.vocab_size, **kwargs)
+
+        aux_loss = None
+        if output_router_logits:
+            aux_loss = load_balancing_loss_func(
+                outputs.router_logits,
+                self.num_experts,
+                self.num_experts_per_tok,
+                attention_mask,
+            )
+            if labels is not None:
+                loss += self.router_aux_loss_coef * aux_loss.to(loss.device)  # make sure to reside in the same device
+
+        return MoeCausalLMOutputWithPast(
+            loss=loss,
+            aux_loss=aux_loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+            router_logits=outputs.router_logits,
+        )
+
+
+class MiniMaxM2ForSequenceClassification(GenericForSequenceClassification, MiniMaxM2PreTrainedModel):
+    pass
+
+
+class MiniMaxM2ForTokenClassification(GenericForTokenClassification, MiniMaxM2PreTrainedModel):
+    pass
+
+
+class MiniMaxM2ForQuestionAnswering(GenericForQuestionAnswering, MiniMaxM2PreTrainedModel):
+    pass
+
+
+__all__ = [
+    "MiniMaxM2ForCausalLM",
+    "MiniMaxM2ForQuestionAnswering",
+    "MiniMaxM2Model",
+    "MiniMaxM2PreTrainedModel",
+    "MiniMaxM2ForSequenceClassification",
+    "MiniMaxM2ForTokenClassification",
+]

tokenizer_config.json

@@ -0,0 +1,495 @@
+{
+  "added_tokens_decoder": {
+  "200000": {
+      "content": "]!p~[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200001": {
+      "content": "<fim_prefix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200002": {
+      "content": "<fim_middle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200003": {
+      "content": "<fim_suffix>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200004": {
+      "content": "<fim_pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200005": {
+      "content": "<reponame>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200006": {
+      "content": "<filename>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200007": {
+      "content": "<gh_stars>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200008": {
+      "content": "<issue_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200009": {
+      "content": "<issue_comment>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200010": {
+      "content": "<issue_closed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200011": {
+    "content": "<jupyter_start>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false,
+    "special": true
+  },
+  "200012": {
+      "content": "<jupyter_text>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200013": {
+      "content": "<jupyter_code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200014": {
+      "content": "<jupyter_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200015": {
+    "content": "<empty_output>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false,
+    "special": true
+  },
+  "200016": {
+      "content": "<commit_before>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200017": {
+      "content": "<commit_msg>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200018": {
+      "content": "<commit_after>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200019": {
+      "content": "]~b]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200020": {
+      "content": "[e~[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200021": {
+      "content": "]!d~[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200022": {
+      "content": "<function_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200023": {
+      "content": "<code_interpreter>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200024": {
+      "content": "]<]speech[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200025": {
+      "content": "]<]image[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200026": {
+      "content": "]<]video[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200027": {
+      "content": "]<]start of speech[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200028": {
+      "content": "]<]end of speech[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200029": {
+      "content": "]<]start of image[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200030": {
+      "content": "]<]end of image[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200031": {
+      "content": "]<]start of video[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200032": {
+      "content": "]<]end of video[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200033": {
+      "content": "]<]vision pad[>[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200034": {
+      "content": "]~!b[",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200035": {
+      "content": "<jupyter_error>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200036": {
+      "content": "<add_file>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": false,
+      "special": true
+  },
+  "200037": {
+      "content": "<delete_file>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200038": {
+      "content": "<rename_file>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200039": {
+      "content": "<edit_file>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200040": {
+      "content": "<commit_message>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200041": {
+      "content": "<empty_source_file>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200042": {
+      "content": "<repo_struct>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+  },
+  "200043": {
+    "content": "<code_context>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": true
+  },
+  "200044": {
+    "content": "<file_content>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": true
+  },
+  "200045": {
+    "content": "<source_files>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": true
+  },
+  "200046": {
+    "content": "<pr_start>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": true
+  },
+  "200047": {
+    "content": "<review_comment>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": true
+  },
+  "200048": {
+    "content": "<filepath>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": true
+  },
+  "200049": {
+    "content": "<file_sep>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": true
+  },
+  "200050": {
+    "content": "<think>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": false
+  },
+  "200051": {
+    "content": "</think>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": false
+  },
+  "200052": {
+    "content": "<minimax:tool_call>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": false
+  },  
+  "200053": {
+    "content": "</minimax:tool_call>",
+    "single_word": false,
+    "lstrip": false,
+    "rstrip": false,
+    "normalized": false,
+    "special": false
+  }
+  },
+  "additional_special_tokens": [
+        "<code_interpreter>",
+        "<commit_after>",
+        "<commit_before>",
+        "<commit_msg>",
+        "<empty_output>",
+        "<filename>",
+        "<fim_middle>",
+        "<fim_pad>",
+        "<fim_prefix>",
+        "<fim_suffix>",
+        "<function_call>",
+        "<gh_stars>",
+        "]<]speech[>[",
+        "]<]image[>[",
+        "]<]video[>[",
+        "]<]start of speech[>[",
+        "]<]end of speech[>[",
+        "]<]start of image[>[",
+        "]<]end of image[>[",
+        "]<]start of video[>[",
+        "]<]end of video[>[",
+        "]<]vision pad[>[",
+        "]~!b[",
+        "<issue_closed>",
+        "<issue_comment>",
+        "<issue_start>",
+        "<jupyter_code>",
+        "<jupyter_output>",
+        "<jupyter_start>",
+        "<jupyter_text>",
+        "<reponame>",
+        "[e~[",
+        "]!d~[",
+        "]!p~[",
+        "]~b]",
+        "<jupyter_error>",
+        "<add_file>",
+        "<delete_file>",
+        "<rename_file>",
+        "<edit_file>",
+        "<commit_message>",
+        "<empty_source_file>",
+        "<repo_struct>",
+        "<code_context>",
+        "<file_content>",
+        "<source_files>",
+        "<pr_start>",
+        "<review_comment>",
+        "<filepath>",
+        "<file_sep>"
+    ],
+  "add_prefix_space": false,
+  "bos_token": "]~!b[",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "[e~[",
+  "model_max_length": 40960000,
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "]!d~["
+}