Instructions to use ModelCloud/DeepSeek-V3-0324-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ModelCloud/DeepSeek-V3-0324-BF16 with Transformers:

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

pipe = pipeline("text-generation", model="ModelCloud/DeepSeek-V3-0324-BF16", 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("ModelCloud/DeepSeek-V3-0324-BF16", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("ModelCloud/DeepSeek-V3-0324-BF16", trust_remote_code=True, device_map="auto")
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 Settings

vLLM

How to use ModelCloud/DeepSeek-V3-0324-BF16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ModelCloud/DeepSeek-V3-0324-BF16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ModelCloud/DeepSeek-V3-0324-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ModelCloud/DeepSeek-V3-0324-BF16

SGLang

How to use ModelCloud/DeepSeek-V3-0324-BF16 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 "ModelCloud/DeepSeek-V3-0324-BF16" \
    --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": "ModelCloud/DeepSeek-V3-0324-BF16",
		"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 "ModelCloud/DeepSeek-V3-0324-BF16" \
        --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": "ModelCloud/DeepSeek-V3-0324-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ModelCloud/DeepSeek-V3-0324-BF16 with Docker Model Runner:
```
docker model run hf.co/ModelCloud/DeepSeek-V3-0324-BF16
```

Qubitium commited on Mar 27, 2025

Commit

be8d061

verified ·

1 Parent(s): 0ed44b2

sync modeling_deepseek.py with upstream

Browse files

Files changed (1) hide show

modeling_deepseek.py +2 -3

modeling_deepseek.py CHANGED Viewed

@@ -398,7 +398,6 @@ class MoEGate(nn.Module):
         self.n_routed_experts = config.n_routed_experts
         self.routed_scaling_factor = config.routed_scaling_factor
         self.scoring_func = config.scoring_func
-        self.seq_aux = config.seq_aux
         self.topk_method = config.topk_method
         self.n_group = config.n_group
         self.topk_group = config.topk_group
@@ -455,7 +454,7 @@ class MoEGate(nn.Module):
                 )
                 .reshape(bsz * seq_len, -1)
             )  # [n, e]
-            tmp_scores = scores_for_choice.masked_fill(~score_mask.bool(), 0.0)  # [n, e]
             _, topk_idx = torch.topk(
                 tmp_scores, k=self.top_k, dim=-1, sorted=False
             )
@@ -1846,4 +1845,4 @@ class DeepseekV3ForSequenceClassification(DeepseekV3PreTrainedModel):
             past_key_values=transformer_outputs.past_key_values,
             hidden_states=transformer_outputs.hidden_states,
             attentions=transformer_outputs.attentions,
-        )

         self.n_routed_experts = config.n_routed_experts
         self.routed_scaling_factor = config.routed_scaling_factor
         self.scoring_func = config.scoring_func
         self.topk_method = config.topk_method
         self.n_group = config.n_group
         self.topk_group = config.topk_group
                 )
                 .reshape(bsz * seq_len, -1)
             )  # [n, e]
+            tmp_scores = scores_for_choice.masked_fill(~score_mask.bool(), float("-inf"))  # [n, e]
             _, topk_idx = torch.topk(
                 tmp_scores, k=self.top_k, dim=-1, sorted=False
             )
             past_key_values=transformer_outputs.past_key_values,
             hidden_states=transformer_outputs.hidden_states,
             attentions=transformer_outputs.attentions,
+        )