Instructions to use deepseek-ai/DeepSeek-V4-Flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use deepseek-ai/DeepSeek-V4-Flash with Transformers:

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

pipe = pipeline("text-generation", model="deepseek-ai/DeepSeek-V4-Flash")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("deepseek-ai/DeepSeek-V4-Flash")
model = AutoModelForCausalLM.from_pretrained("deepseek-ai/DeepSeek-V4-Flash")

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use deepseek-ai/DeepSeek-V4-Flash with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "deepseek-ai/DeepSeek-V4-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": "deepseek-ai/DeepSeek-V4-Flash",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/deepseek-ai/DeepSeek-V4-Flash

SGLang

How to use deepseek-ai/DeepSeek-V4-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 "deepseek-ai/DeepSeek-V4-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": "deepseek-ai/DeepSeek-V4-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 "deepseek-ai/DeepSeek-V4-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": "deepseek-ai/DeepSeek-V4-Flash",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use deepseek-ai/DeepSeek-V4-Flash with Docker Model Runner:
```
docker model run hf.co/deepseek-ai/DeepSeek-V4-Flash
```

bsarpel

#14

by bsarpel - opened 13 days ago

base: refs/heads/main

←

from: refs/pr/14

Discussion Files changed

-4

This PR is in draft mode

Files changed (3) hide show

DeepSeek_V4.pdf +2 -2
config.json +0 -1
inference/model.py +2 -1

DeepSeek_V4.pdf CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8a03dadae71894de1515da33e296f0df1dbeed3e7f4bf0badd087f9af77f29e9
-size 4480407

 version https://git-lfs.github.com/spec/v1
+oid sha256:f4cbe4fcbd2888b25b2890a98cc6ef4ce0489df7c93e140b6f853c451d3f5c52
+size 4479907

config.json CHANGED Viewed

@@ -6,7 +6,6 @@
   "attention_dropout": 0.0,
   "bos_token_id": 0,
   "eos_token_id": 1,
-  "expert_dtype": "fp4",
   "hc_eps": 1e-06,
   "hc_mult": 4,
   "hc_sinkhorn_iters": 20,

   "attention_dropout": 0.0,
   "bos_token_id": 0,
   "eos_token_id": 1,
   "hc_eps": 1e-06,
   "hc_mult": 4,
   "hc_sinkhorn_iters": 20,

inference/model.py CHANGED Viewed

@@ -624,7 +624,8 @@ class MoE(nn.Module):
         self.experts = nn.ModuleList([Expert(args.dim, args.moe_inter_dim, dtype=expert_dtype, swiglu_limit=args.swiglu_limit) if self.experts_start_idx <= i < self.experts_end_idx else None
                                        for i in range(self.n_routed_experts)])
         assert args.n_shared_experts == 1
-        self.shared_experts = Expert(args.dim, args.moe_inter_dim, swiglu_limit=args.swiglu_limit)
     def forward(self, x: torch.Tensor, input_ids: torch.Tensor) -> torch.Tensor:
         shape = x.size()

         self.experts = nn.ModuleList([Expert(args.dim, args.moe_inter_dim, dtype=expert_dtype, swiglu_limit=args.swiglu_limit) if self.experts_start_idx <= i < self.experts_end_idx else None
                                        for i in range(self.n_routed_experts)])
         assert args.n_shared_experts == 1
+        # no swiglu_limit
+        self.shared_experts = Expert(args.dim, args.moe_inter_dim)
     def forward(self, x: torch.Tensor, input_ids: torch.Tensor) -> torch.Tensor:
         shape = x.size()