Nanbeige
/

Nanbeige4.2-3B

Text Generation

Model card Files Files and versions

Instructions to use Nanbeige/Nanbeige4.2-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Nanbeige/Nanbeige4.2-3B with Transformers:

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

pipe = pipeline("text-generation", model="Nanbeige/Nanbeige4.2-3B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Nanbeige/Nanbeige4.2-3B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Nanbeige/Nanbeige4.2-3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Nanbeige/Nanbeige4.2-3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Nanbeige/Nanbeige4.2-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Nanbeige/Nanbeige4.2-3B

How to use Nanbeige/Nanbeige4.2-3B 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 "Nanbeige/Nanbeige4.2-3B" \
    --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": "Nanbeige/Nanbeige4.2-3B",
		"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 "Nanbeige/Nanbeige4.2-3B" \
        --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": "Nanbeige/Nanbeige4.2-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Nanbeige/Nanbeige4.2-3B with Docker Model Runner:
```
docker model run hf.co/Nanbeige/Nanbeige4.2-3B
```

Resources

View closed (5)

Very Excellent Model!!!!!!!!!

#36 opened 6 days ago by

Q4_K_M slower than Q8_0

#35 opened 7 days ago by

Any plans to get this working with openvino?

#34 opened 23 days ago by

Such an interesting model

#33 opened 25 days ago by

I'm curious, what tasks would this kind of model be used for? Or is it purely for research purposes?

#32 opened about 1 month ago by

Why is it much slower and much more memory hungry than slightly larger models such as Qwen3.5-4B?

#31 opened about 1 month ago by

slightlyoutofphase

GPQA复现上出现问题

#30 opened about 1 month ago by

General Release Date for 4.5?

#29 opened about 1 month ago by

Long context inference

#28 opened about 1 month ago by

Any path to getting this integrated into the latest docker build of llama.cpp?

#27 opened about 1 month ago by

Pretrain recipe questions for small-model compute budgeting (not data)

#26 opened about 1 month ago by

Tested on the Artificial Analysis Intelligence Index?

#25 opened about 1 month ago by

Awesome!

#24 opened about 2 months ago by

llama-server not in build files

#23 opened about 2 months ago by

web search

#22 opened about 2 months ago by

Amazing for its size, but spirals into questionable solutions once it fails something

#21 opened about 2 months ago by

Massive activation at layer 21, and three follow-ups on loop depth + LoopSplit

#20 opened about 2 months ago by

Suitability for 8 GB graphics cards

#18 opened about 2 months ago by

llama.cpp silently drops valid tool calls (parser bug, not the model) - 2-line fix for your fork!

#17 opened about 2 months ago by

Undocumented config params + two questions on loop depth and length-control RL

#16 opened about 2 months ago by

Nanbeige4.2-3B: A Small Model That Demands Serious Hardware

#15 opened about 2 months ago by

Some questions about training

#14 opened about 2 months ago by

Installation Video and Testing - Step by Step

#13 opened about 2 months ago by

Do you have any plans to include multilingual support?

#12 opened about 2 months ago by

preserve_thinking

#9 opened about 2 months ago by

Please specify max context size in model's card.

#8 opened about 2 months ago by

Issue: Grammar-constrained decoding (`response_format: json_schema`) fails at sampler init — but `--no-jinja` fixes it

#6 opened about 2 months ago by

Any plans for FP8?

#5 opened about 2 months ago by

Q on the model

#3 opened about 2 months ago by

damn i was in the process of doing something very similar

#2 opened about 2 months ago by

gguf

#1 opened about 2 months ago by