Instructions to use t-tech/T-pro-it-2.0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use t-tech/T-pro-it-2.0 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="t-tech/T-pro-it-2.0") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("t-tech/T-pro-it-2.0") model = AutoModelForCausalLM.from_pretrained("t-tech/T-pro-it-2.0") 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]:])) - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use t-tech/T-pro-it-2.0 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "t-tech/T-pro-it-2.0" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "t-tech/T-pro-it-2.0", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/t-tech/T-pro-it-2.0
- SGLang
How to use t-tech/T-pro-it-2.0 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 "t-tech/T-pro-it-2.0" \ --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": "t-tech/T-pro-it-2.0", "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 "t-tech/T-pro-it-2.0" \ --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": "t-tech/T-pro-it-2.0", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use t-tech/T-pro-it-2.0 with Docker Model Runner:
docker model run hf.co/t-tech/T-pro-it-2.0
Ошибка при развертывании LLM: не инициализированы веса из чекпоинта
Привет всем!
При попытке развернуть модель LLM, столкнулся с ошибкой, связанной с тем, что часть весов не была загружена из чекпоинта.
ValueError: Following weights were not initialized from checkpoint: {'model.layers.58.post_attention_layernorm.weight', 'model.layers.63.self_attn.k_norm.weight', 'model.layers.61.self_attn.q_norm.weight', 'model.layers.58.mlp.down_proj.weight', 'model.layers.62.self_attn.q_norm.weight', 'model.layers.62.self_attn.k_norm.weight', 'model.layers.62.mlp.gate_up_proj.weight', 'model.layers.60.self_attn.o_proj.weight', 'model.layers.60.mlp.down_proj.weight', 'model.layers.60.mlp.gate_up_proj.weight', 'model.layers.59.post_attention_layernorm.weight', 'model.layers.63.self_attn.qkv_proj.weight', 'model.layers.60.self_attn.q_norm.weight', 'model.layers.61.input_layernorm.weight', 'model.layers.59.self_attn.k_norm.weight', 'model.layers.61.self_attn.o_proj.weight', 'model.layers.62.post_attention_layernorm.weight', 'model.layers.61.post_attention_layernorm.weight', 'model.layers.59.mlp.down_proj.weight', 'model.layers.61.self_attn.k_norm.weight', 'model.layers.61.mlp.gate_up_proj.weight', 'model.layers.62.self_attn.o_proj.weight', 'model.layers.63.self_attn.o_proj.weight', 'model.layers.60.input_layernorm.weight', 'model.layers.58.input_layernorm.weight', 'model.layers.59.self_attn.o_proj.weight', 'model.layers.60.self_attn.qkv_proj.weight', 'model.layers.62.input_layernorm.weight', 'model.layers.60.self_attn.k_norm.weight', 'model.layers.60.post_attention_layernorm.weight', 'model.layers.59.mlp.gate_up_proj.weight', 'model.layers.59.input_layernorm.weight', 'model.layers.62.self_attn.qkv_proj.weight', 'model.layers.61.self_attn.qkv_proj.weight', 'model.layers.62.mlp.down_proj.weight', 'model.layers.59.self_attn.qkv_proj.weight', 'model.layers.59.self_attn.q_norm.weight', 'model.layers.61.mlp.down_proj.weight', 'model.layers.63.self_attn.q_norm.weight'}
Может быть, кто-то сталкивался с похожей проблемой? Возможно, это связано с несоответствием версий модели и кода, или с тем, что чекпоинт был неполным?
Используемая модель: T-pro-it-2.0
Фреймворк: vLLM 0.8.5.post1
Буду благодарен за любую помощь или указание на возможную причину проблемы.
А какая видеокарта стоит? Там относительно видюхи надо разные Версии vLLM ставить. Если "пользовательские" видеокарты по типу 3090/4090, то лучше разворачивать на SGLang, они лучше адаптированы под них.
Карточка Nvidia H100