Instructions to use vilm/vinallama-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use vilm/vinallama-7b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="vilm/vinallama-7b")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("vilm/vinallama-7b") model = AutoModelForCausalLM.from_pretrained("vilm/vinallama-7b") - Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use vilm/vinallama-7b with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "vilm/vinallama-7b" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "vilm/vinallama-7b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/vilm/vinallama-7b
- SGLang
How to use vilm/vinallama-7b 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 "vilm/vinallama-7b" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "vilm/vinallama-7b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'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 "vilm/vinallama-7b" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "vilm/vinallama-7b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use vilm/vinallama-7b with Docker Model Runner:
docker model run hf.co/vilm/vinallama-7b
Gặp vấn đề khi finetune
Mình finetune mô hình thì gặp lỗi này:
'''
trainer = transformers.Trainer(
model = model,
train_dataset = data,
args = training_args,
data_collator = transformers.DataCollatorForLanguageModeling(tokenizer, mlm = False),
)
model.config.use_cache = False
trainer.train()
'''
ValueError: You cannot perform fine-tuning on purely quantized models. Please attach trainable adapters on top of the quantized model to correctly perform fine-tuning. Please see: https://huggingface.co/docs/transformers/peft for more details
Theo mình hiểu thì khi bạn fine-tune theo kiểu dùng QLoRa hay LoRa thì phải gửi kèm cái adapter thì mới train được. Do khi train model trong Deep learning ngta sẽ sử dụng kiểu dữ liệu float32(hoặc float16) còn khi inference mới quantize xuống để dùng, nên việc fine-tune trên model đã quantize là không đúng á (Ý kiến cá nhân)