Instructions to use amd/MiniMax-M2.7-MXFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use amd/MiniMax-M2.7-MXFP4 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="amd/MiniMax-M2.7-MXFP4", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("amd/MiniMax-M2.7-MXFP4", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("amd/MiniMax-M2.7-MXFP4", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use amd/MiniMax-M2.7-MXFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "amd/MiniMax-M2.7-MXFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "amd/MiniMax-M2.7-MXFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/amd/MiniMax-M2.7-MXFP4

SGLang

How to use amd/MiniMax-M2.7-MXFP4 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "amd/MiniMax-M2.7-MXFP4" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "amd/MiniMax-M2.7-MXFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "amd/MiniMax-M2.7-MXFP4" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "amd/MiniMax-M2.7-MXFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use amd/MiniMax-M2.7-MXFP4 with Docker Model Runner:
```
docker model run hf.co/amd/MiniMax-M2.7-MXFP4
```

haoyang-amd commited on Apr 23

Commit

ba4fa61

verified ·

1 Parent(s): f1c5e1e

Update README.md

Browse files

Files changed (1) hide show

README.md +39 -10

README.md CHANGED Viewed

@@ -14,9 +14,9 @@ license_link: https://huggingface.co/MiniMaxAI/MiniMax-M2.7/blob/main/LICENSE
   - **Input:** Text
   - **Output:** Text
 - **Supported Hardware Microarchitecture:** AMD MI300 MI350/MI355
-- **ROCm**: ---
-- **PyTorch**: ---
-- **Transformers**: ---
 - **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)
@@ -26,16 +26,29 @@ license_link: https://huggingface.co/MiniMaxAI/MiniMax-M2.7/blob/main/LICENSE
 # Model Quantization
-The model was quantized from [MiniMaxAI/MiniMax-M2.7](https://huggingface.co/MiniMaxAI/MiniMax-M2.7) using [AMD-Quark](https://quark.docs.amd.com/latest/index.html). The weights are quantized to MXFP4 and activations are quantized to MXFP4.
 **Quantization scripts:**
-TBD
 For further details or issues, please refer to the AMD-Quark documentation or contact the respective developers.
 # Evaluation
-TBD
 ### Accuracy
@@ -53,11 +66,11 @@ TBD
   <tr>
    <td>gsm8k (flexible-extract)
    </td>
-   <td>TBD
    </td>
-   <td>TBD
    </td>
-   <td>TBD
    </td>
   </tr>
 </table>
@@ -65,7 +78,23 @@ TBD
 ### Reproduction
-TBD
 # License
 Modifications Copyright(c) 2026 Advanced Micro Devices, Inc. All rights reserved.

   - **Input:** Text
   - **Output:** Text
 - **Supported Hardware Microarchitecture:** AMD MI300 MI350/MI355
+- **ROCm**: 7.2.1
+- **PyTorch**: 2.10.0+git8514f05
+- **Transformers**: 4.57.1
 - **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)
 # Model Quantization
+The model was quantized from [amd/MiniMax-M2.7-BF16](https://huggingface.co/amd/MiniMax-M2.7-BF16) using [AMD-Quark](https://quark.docs.amd.com/latest/index.html). The weights are quantized to MXFP4 and activations are quantized to MXFP4.
 **Quantization scripts:**
+```
+export exclude_layers="lm_head *block_sparse_moe.gate* *self_attn*"
+export HIP_VISIBLE_DEVICES=4,5,6,7
+python3 quantize_quark.py --model_dir /shareddata/amd/MiniMax-M2.7-bf16 \
+                          --quant_scheme mxfp4 \
+                          --num_calib_data 128 \
+                          --exclude_layers $exclude_layers \
+                          --skip_evaluation \
+                          --multi_gpu  \
+                          --trust_remote_code \
+                          --model_export hf_format \
+                          --output_dir /shareddata/amd/MiniMax-M2.7-mxfp4
+```
 For further details or issues, please refer to the AMD-Quark documentation or contact the respective developers.
 # Evaluation
+The model was evaluated on gsm8k benchmarks using the vllm framework.
 ### Accuracy
   <tr>
    <td>gsm8k (flexible-extract)
    </td>
+   <td>91.36
    </td>
+   <td>91.81
    </td>
+   <td>100.49%
    </td>
   </tr>
 </table>
 ### Reproduction
+The GSM8K results were obtained using the lm-eval framework, based on the Docker image `rocm/vllm-dev:nightly`.
+#### Launching server
+```
+vllm serve "$MODEL" \
+    --tensor-parallel-size 4 \
+    --trust-remote-code \
+    --max-model-len 32768 \
+    --port 8899
+```
+#### Evaluating model in a new terminal
+```
+python vllm/tests/evals/gsm8k/gsm8k_eval.py --host http://127.0.0.1 --port 8899
+```
 # License
 Modifications Copyright(c) 2026 Advanced Micro Devices, Inc. All rights reserved.