Instructions to use boqiny/Qwen3-8B-FengGe-SFT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use boqiny/Qwen3-8B-FengGe-SFT with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="boqiny/Qwen3-8B-FengGe-SFT") 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("boqiny/Qwen3-8B-FengGe-SFT") model = AutoModelForCausalLM.from_pretrained("boqiny/Qwen3-8B-FengGe-SFT", 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 boqiny/Qwen3-8B-FengGe-SFT with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "boqiny/Qwen3-8B-FengGe-SFT" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "boqiny/Qwen3-8B-FengGe-SFT", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/boqiny/Qwen3-8B-FengGe-SFT
- SGLang
How to use boqiny/Qwen3-8B-FengGe-SFT 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 "boqiny/Qwen3-8B-FengGe-SFT" \ --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": "boqiny/Qwen3-8B-FengGe-SFT", "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 "boqiny/Qwen3-8B-FengGe-SFT" \ --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": "boqiny/Qwen3-8B-FengGe-SFT", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use boqiny/Qwen3-8B-FengGe-SFT with Docker Model Runner:
docker model run hf.co/boqiny/Qwen3-8B-FengGe-SFT
Qwen3-8B-FengGe-SFT
๐ Overview
Qwen3-8B-FengGe-SFT is a LoRA fine-tuned conversational model based on Qwen/Qwen3-8B, trained to mimic the distinctive speaking style (โๅณฐๅฅโ) using the Zhoulifeng Streaming Dataset.
๐ Training data: https://huggingface.co/datasets/hzb29/Zhoulifeng-Streaming-Dataset
The model is designed for informal, conversational Chinese dialogue with a strong personality style, emphasizing long-form, free-flowing responses.
๐ง Base Model
- Model:
Qwen/Qwen3-8B - Fine-tuning method: LoRA
๐ Training Loss
๐ Inference
Load with Hugging Face
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "boqiny/Qwen3-8B-FengGe-SFT"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)
prompt = "ๅณฐๅฅๆไผผไนๆปๆฏ็ฑไธไธไผๅๆฌขๆ็ๅฅณ็๏ผๆ็ๅนฒ่ไธๆญ็๏ผ่ฟๆฏๅฝ่ฟ็ๅฎๆ่ฟๆฏๆงๅๆ"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(
**inputs,
max_new_tokens=256,
repetition_penalty=1.3
)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
โ ๏ธ Note:
repetition_penalty=1.3is important to reduce looping/repetitive outputs.
๐ฌ Example Generations
Example 1
User:
ๅณฐๅฅๆไผผไนๆปๆฏ็ฑไธไธไผๅๆฌขๆ็ๅฅณ็๏ผๆ็ๅนฒ่ไธๆญ็๏ผ่ฟๆฏๅฝ่ฟ็ๅฎๆ่ฟๆฏๆงๅๆ
Assistant:
ไฝ ้ฝๆฒกๆๆฅ่งฆ่ฟ็ๅฏน่ฑกไฝ ๆไนๅฐฑ็ฅ้ไบบๅฎถ่ฎจๅไฝ ไนๅๆฌขไธไธไฝ ่ชๅทฑๅ๏ผๅฐฑๆฏๅ ไธบไธ็ฅ้ๅๅ ๆ่ฝ่ฎฉๅคง่่กๆไนฑๆณไบใๅๆญฃไฝ ่ฆๆฏๆๆบไผ็ๆญฃ่ง้ข่ฏ่ฏๆ็ฅ้ๅ
ทไฝ็็ปๆๅฐๅบๆฏๆไนๅไบใๆๆถๅ็ท็ๅจไธ่ตทๅๅ่่ๅคฉๆ็ๆๆฃๆญฅ็ตๆๆ่
ฐๆฑๅคง่
ฟ่ทณ่ๅฑๆญๅ้
ๅ้ฅญ้่ก้ๅๅบ็็ตๅฝฑๅๅพฎไฟกๅทๆ้ณ่ง้ข็ดๆญๆๆธธๆๅฅ่บซ้ป็ผ่ฟๅจ็ปๅฑฑ็ฌๅฑฑ้ฒ่ฅ่นฆ่ฟชๅนฟๅบ่ก่พน้ฒๆธธๅคๅบ้
ๅงๆกๆฟๆๆฉ่ถณ็KTVๆญๅ
้
ๅบๆฐๅฎฟๅๅฎถไน็ง็คๅญๆธธๆณณ้ฆๆธฉๆณๆฑ ๆฐดไนๅญๅจ็ฉๅญๆธธไนๅบๅ
ฌๅญ่ฑๅญ็ๅบ่ถณ็ๅบ็ฏฎ็ๅบ็พฝๆฏ็ๅบ็ฝ็ๅบๅฐ่ช่ก่ฝฆ่ต้้ฉฌ็่ตๅฐ็ฎญ้ถๆ ๅฐๅปๅบ้ฆๆๆๆ ผๆก็ซๆ่งๆณๅปๆ่ทค็ฉบๆๅฏนๆ่กจๆผ้ญๅนป้ญๆฏๆๆ็น่บ็ญ้ฝ่ฝไฝ้ชไธๆฌกไธคๅฃๅญๅจไธ่ตทๆถๅฟซไนๆถๅ
่ถ้ฟ่ถๅฅฝ๏ผๅจไธๅ็ๆถ้ด่ถๅคๅฐฑ่ถไธๅฎนๆๅๅผ็ฆปๅผๅฝผๆญคใ็ธๅๅฆๆไฝ ไธ็ดๅ้ฃๅญคๅๅฅๆ็ฌ่ชไธไบบๅฝขๅๅฝฑๅชๅญค็ฌ็ป่ไธ่พๅญไนๆฒก่ฐๆๆ็ฑๆไนๅๅข๏ผ่ฟ่ฝๅนฒ็นๅซ็ไบๆบๅผๅฟ็๏ผๆพไธ็พค็ทๅฎถไบบ็ปๅขๅบๅป็ฉไธๅไนๅพไธ้ใ
Example 2
User:
ๅณฐๅฅ๏ผไฝ ๆฏๆไนๅๅฐ่งฃ็ญไธ้ดไธ็ฉๆๅๅพๅบ่ฟๆฏๅฅฝไบ็็ป่ฎบ๏ผ่ฝๆๆๆๅ
Assistant:
้ฃๅฐฑๆฏไฝ ็ไปทๅผ่งๅฐฑๆฏ่ฟๆ ทๅญ่ๅทฒใๆไบไบบๅฏ่ฝ่ฎคไธบ่ฟๆ ทไธๅฅฝ้ฃๆ ทไธ่ก๏ผไฝๆ่งๅพๅฅฝ็ๅฐๆนๅคชๅคไบใๅช่ฆ่ฝๅคๆปก่ถณไบบ็ฑป็ๅบๆฌ้ๆฑๅฐฑๅฏไปฅ๏ผๆฒกๅฟ
่ฆ่ฟฝๆฑไธไบ้ซ็ฒพๅฐ็ไธ่ฅฟ๏ผๆฒกๆๅฟ
่ฆๅ้ฃ็ง่ถ
่ถๆถไปฃ็ไบๅฟไนไธๅฟ
่ฆๆไธบๅไบบๅฏ่ฑช็งๅญฆๅฎถๅฅ็๏ผๅช่ฆๆฏๅฎๅฎ็จณ็จณ็็ๆดปๅฐฑๅฅฝใๆไปฅ่ฏดๆ็่ง็นไนๆฏๅบไบๆ่ชๅทฑ็ๆดป็ป้ชไนไธๅฝขๆ็่ฟไนไธช็ๆณใๆฒกๆไปไน็นๅซ็ๅๅ ๅชๆฏๅ ไธบๆๆดๅๆฌข่ฟๆ ท็็ๆดปๆนๅผ็ฝขไบ๏ผๅฐฑ่ฟไน็ฎๅใ
๐ฆ Training with LLaMA-Factory
This model was trained using LLaMA-Factory with the following configuration:
### model
model_name_or_path: Qwen/Qwen3-8B
trust_remote_code: true
### method
stage: sft
do_train: true
finetuning_type: lora
lora_rank: 32
lora_target: all
### dataset
dataset: zhoulifeng_streaming
template: qwen3_nothink
cutoff_len: 2048
### train
per_device_train_batch_size: 1
gradient_accumulation_steps: 8
learning_rate: 5.0e-5
num_train_epochs: 2.0
lr_scheduler_type: cosine
warmup_ratio: 0.1
bf16: true
โ ๏ธ Limitations
- Strong stylistic bias toward โๅณฐๅฅโ conversational tone
- May generate long, repetitive enumerations without proper decoding settings
- Not suitable for factual QA or safety-critical applications
๐ License
Please follow the license of the base model:
๐ Qwen/Qwen3-8B
- Downloads last month
- 10
