Instructions to use kshitijthakkar/deepseek-v4-mini-300M-init with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use kshitijthakkar/deepseek-v4-mini-300M-init with Transformers:

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

pipe = pipeline("text-generation", model="kshitijthakkar/deepseek-v4-mini-300M-init")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("kshitijthakkar/deepseek-v4-mini-300M-init")
model = AutoModelForCausalLM.from_pretrained("kshitijthakkar/deepseek-v4-mini-300M-init")
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 kshitijthakkar/deepseek-v4-mini-300M-init with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "kshitijthakkar/deepseek-v4-mini-300M-init"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kshitijthakkar/deepseek-v4-mini-300M-init",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/kshitijthakkar/deepseek-v4-mini-300M-init

SGLang

How to use kshitijthakkar/deepseek-v4-mini-300M-init 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 "kshitijthakkar/deepseek-v4-mini-300M-init" \
    --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": "kshitijthakkar/deepseek-v4-mini-300M-init",
		"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 "kshitijthakkar/deepseek-v4-mini-300M-init" \
        --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": "kshitijthakkar/deepseek-v4-mini-300M-init",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use kshitijthakkar/deepseek-v4-mini-300M-init with Docker Model Runner:
```
docker model run hf.co/kshitijthakkar/deepseek-v4-mini-300M-init
```

deepseek-v4-mini-300M-init / code /deepseek_v4 /__init__.py

kshitijthakkar

Add modeling+configuration package for Colab validation

4d4e476 verified 7 days ago

raw

history blame contribute delete

1.17 kB

	"""DeepSeek-V4 architecture package.

	Importing this module registers ``DeepseekV4Config`` / ``DeepseekV4ForCausalLM``
	with the HuggingFace ``transformers`` auto classes so that

	from transformers import AutoModelForCausalLM
	AutoModelForCausalLM.from_pretrained("path/to/our/checkpoint")

	resolves to ``DeepseekV4ForCausalLM`` without ``trust_remote_code=True``.
	"""
	from transformers import AutoConfig, AutoModel, AutoModelForCausalLM

	from .configuration_deepseek_v4 import DeepseekV4Config
	from .modeling_deepseek_v4 import (
	DeepseekV4Model,
	DeepseekV4ForCausalLM,
	DeepseekV4PreTrainedModel,
	)

	# Idempotent registration — safe to import twice.
	try:
	AutoConfig.register("deepseek_v4", DeepseekV4Config, exist_ok=True)
	except (ValueError, TypeError):
	pass
	try:
	AutoModel.register(DeepseekV4Config, DeepseekV4Model, exist_ok=True)
	except (ValueError, TypeError):
	pass
	try:
	AutoModelForCausalLM.register(DeepseekV4Config, DeepseekV4ForCausalLM, exist_ok=True)
	except (ValueError, TypeError):
	pass

	__all__ = [
	"DeepseekV4Config",
	"DeepseekV4Model",
	"DeepseekV4ForCausalLM",
	"DeepseekV4PreTrainedModel",
	]