Instructions to use XiaomiMiMo/MiMo-V2-Flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use XiaomiMiMo/MiMo-V2-Flash with Transformers:

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

pipe = pipeline("text-generation", model="XiaomiMiMo/MiMo-V2-Flash", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("XiaomiMiMo/MiMo-V2-Flash", trust_remote_code=True, dtype="auto")

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use XiaomiMiMo/MiMo-V2-Flash with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "XiaomiMiMo/MiMo-V2-Flash"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "XiaomiMiMo/MiMo-V2-Flash",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/XiaomiMiMo/MiMo-V2-Flash

SGLang

How to use XiaomiMiMo/MiMo-V2-Flash 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 "XiaomiMiMo/MiMo-V2-Flash" \
    --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": "XiaomiMiMo/MiMo-V2-Flash",
		"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 "XiaomiMiMo/MiMo-V2-Flash" \
        --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": "XiaomiMiMo/MiMo-V2-Flash",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use XiaomiMiMo/MiMo-V2-Flash with Docker Model Runner:
```
docker model run hf.co/XiaomiMiMo/MiMo-V2-Flash
```

[Fix.] apply attention_value_scale.

#39

by Jumbo0715 - opened Apr 20

base: refs/heads/main

←

from: refs/pr/39

Discussion Files changed

-0

Files changed (1) hide show

modeling_mimo_v2_flash.py +5 -0

modeling_mimo_v2_flash.py CHANGED Viewed

@@ -330,6 +330,8 @@ class MiMoV2Attention(nn.Module):
         self.v_proj = nn.Linear(config.hidden_size, v_hidden_size, bias=self.attention_bias)
         self.o_proj = nn.Linear(o_hidden_size, config.hidden_size, bias=False)
         self.attention_sink_bias = (
             torch.nn.Parameter(torch.empty(config.num_attention_heads), requires_grad=False)
             if (config.add_full_attention_sink_bias and not is_swa) or (config.add_swa_attention_sink_bias and is_swa)
@@ -353,6 +355,9 @@ class MiMoV2Attention(nn.Module):
         query_states = self.q_proj(hidden_states).view(qk_hidden_shape).transpose(1, 2)
         key_states = self.k_proj(hidden_states).view(qk_hidden_shape).transpose(1, 2)
         value_states = self.v_proj(hidden_states).view(v_hidden_shape).transpose(1, 2)
         cos, sin = position_embeddings

         self.v_proj = nn.Linear(config.hidden_size, v_hidden_size, bias=self.attention_bias)
         self.o_proj = nn.Linear(o_hidden_size, config.hidden_size, bias=False)
+        self.v_scale = getattr(config, "attention_value_scale", None)
         self.attention_sink_bias = (
             torch.nn.Parameter(torch.empty(config.num_attention_heads), requires_grad=False)
             if (config.add_full_attention_sink_bias and not is_swa) or (config.add_swa_attention_sink_bias and is_swa)
         query_states = self.q_proj(hidden_states).view(qk_hidden_shape).transpose(1, 2)
         key_states = self.k_proj(hidden_states).view(qk_hidden_shape).transpose(1, 2)
         value_states = self.v_proj(hidden_states).view(v_hidden_shape).transpose(1, 2)
+        if self.v_scale is not None:
+            value_states = value_states * self.v_scale
         cos, sin = position_embeddings