LLM入门3 | 基于cpu和hugging face的LLaMA部署

from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("decapoda-research/llama-7b-hf")
model = AutoModelForCausalLM.from_pretrained("decapoda-research/llama-7b-hf")

from huggingface_hub import snapshot_download
snapshot_download(repo_id="decapoda-research/llama-7b-hf")

# 安装transformers
pip install git+https://github.com/zphang/transformers@llama_push

{"bos_token": "", "eos_token": "", "model_max_length": 1000000000000000019884624838656, "tokenizer_class": "LlamaTokenizer", "unk_token": ""}

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained(llama_7b_localpath)

model = AutoModelForCausalLM.from_pretrained(llama_7b_localpath)

from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained(llama_7b_localpath)
model = AutoModelForCausalLM.from_pretrained(llama_7b_localpath)
inp = tokenizer("The capital of Chine?",return_tensors="pt")
outp = model.generate(inp.input_ids)
tokenizer.batch_decode(outp, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]