Instructions to use UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B with Transformers:

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

pipe = pipeline("text-generation", model="UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B")
model = AutoModelForCausalLM.from_pretrained("UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B

SGLang

How to use UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B 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 "UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B" \
    --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": "UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B",
		"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 "UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B" \
        --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": "UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B with Docker Model Runner:
```
docker model run hf.co/UNIVA-Bllossom/DeepSeek-qwen-Bllossom-32B
```

Request for tokenizer.model File for DeepSeek-qwen-Bllossom-32B

by WOOYT99 - opened Jul 7, 2025

Discussion

WOOYT99

Jul 7, 2025

Hello,

I’m currently working with the model DeepSeek-qwen-Bllossom-32B, which appears to be based on the Qwen tokenizer architecture. However, the tokenizer.model file is missing in the current distribution, and I’m encountering errors when trying to load the tokenizer with use_fast=False.

Could you please provide the corresponding tokenizer.model file or let me know the best source from which I can obtain a compatible version?

This is critical for enabling sentencepiece-based tokenization in a local environment where fast tokenizers are not applicable.

Thank you very much in advance!

jerry3972

UNIVA and KAIST-MLP lab org Jul 10, 2025

Hello,

I’ve now added the merges.txt and vocab.json files to the DeepSeek-qwen-Bllossom-32B release, so you can load the tokenizer with use_fast=False without any errors

Please let me know if you run into any further issues!

Thank you.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment