README.md

---
base_model:
- MiniMaxAI/MiniMax-M2.5
language:
- en
library_name: transformers
license: other
license_name: modified-mit
license_link: https://github.com/MiniMax-AI/MiniMax-M2.5/blob/main/LICENSE
---

# Model Overview

- **Model Architecture:** MiniMaxM2ForCausalLM
  - **Input:** Text
  - **Output:** Text
- **Supported Hardware Microarchitecture:** AMD MI300/MI350/MI355 (emulation)
- **ROCm:** 7.2.2
- **PyTorch**: 2.10.0
- **Transformers**: 5.2.0
- **Operating System(s):** Linux
- **Inference Engine:** [SGLang](https://docs.sglang.ai/)/[vLLM](https://docs.vllm.ai/en/latest/)
- **Model Optimizer:** [AMD-Quark](https://quark.docs.amd.com/latest/index.html) (v0.12)
  - **Quantized layers:** `experts`
  - **Weight quantization:** NVFP4, Static 
  - **Activation quantization:** NVFP4, Dynamic


# Model Quantization

The model was quantized from [MiniMaxAI/MiniMax-M2.5](https://huggingface.co/MiniMaxAI/MiniMax-M2.5) by using [AMD-Quark](https://quark.docs.amd.com/latest/index.html). The weights and activations are quantized to NVFP4.


**Quantization scripts:**
```
cd Quark/examples/torch/language_modeling/llm_ptq/
export exclude_layers="lm_head *block_sparse_moe.gate* *self_attn*"
export CUDA_VISIBLE_DEVICES=0,1,2,3
python3 quantize_quark.py \
  --model_dir MiniMaxAI/MiniMax-M2.5 \
  --quant_scheme nvfp4 \
  --num_calib_data 128 \
  --exclude_layers $exclude_layers \
  --model_export hf_format  \
  --trust_remote_code \
  --multi_gpu \
  --output_dir amd/MiniMax-M2.5-NVFP4

```
For further details or issues, please refer to the AMD-Quark documentation or contact the respective developers.

# Deployment
## Use with vLLM/SGLang

This model can be deployed efficiently using the [vLLM](https://docs.vllm.ai/en/latest/) and [SGLang](https://docs.sglang.ai/) backends.

## Evaluation
The model was evaluated on gsm8k benchmarks using the [vllm](https://github.com/vllm-project/vllm/tree/v0.13.0) framework.

### Accuracy

<table>
  <tr>
   <td><strong>Benchmark</strong>
   </td>
   <td><strong>MiniMaxAI/MiniMax-M2.5 </strong>
   </td>
   <td><strong>amd/MiniMax-M2.5-NVFP4(this model)</strong>
   </td>
   <td><strong>Recovery</strong>
   </td>
  </tr>
  <tr>
   <td>gsm8k (flexible-extract) 
   </td>
   <td>91.51
   </td>
   <td>91.21
   </td>
   <td>99.67%
   </td>
  </tr>
</table>



### Reproduction

The GSM8K result was obtained using the `lm-evaluation-harness` framework, based on the Docker image `rocm/vllm-dev:nightly_main_20260603`.

Install the lm-eval `(Version: 0.4.12)` in container first.
```
pip install lm-eval
pip install lm-eval[api]
```
#### Launching server
```
VLLM_ROCM_USE_AITER=1 vllm serve amd/MiniMax-M2.5-NVFP4/ \
  --tensor-parallel-size 2 \
  --tool-call-parser minimax_m2 \
  --reasoning-parser minimax_m2 \
  --enable-auto-tool-choice \
  --trust-remote-code
```
#### Evaluating model in a new terminal
```
lm_eval \
  --model local-completions \
  --model_args "model=amd/MiniMax-M2.5-NVFP4/,base_url=http://127.0.0.1:8000/v1/completions,tokenized_requests=False,tokenizer_backend=None,num_concurrent=32" \
  --gen_kwargs temperature=1.0,top_p=0.95 \
  --tasks gsm8k \
  --num_fewshot 8 \
  --batch_size 1
```


# License
Modifications Copyright(c) 2026 Advanced Micro Devices, Inc. All rights reserved.