生物医药相关大模型：BioMedLM、Mixtral_BioMedical、Med-PaLM2、BioMistral、BioMedGPT

BioMedLM：
https://huggingface.co/stanford-crfm/BioMedLM
Mixtral_BioMedical：
https://huggingface.co/LeroyDyer/Mixtral_BioMedical_7b
Med-PaLM：
https://sites.research.google/med-palm/
BioMistral：
https://huggingface.co/BioMistral/BioMistral-7B
BioMedGPT：
https://github.com/PharMolix/OpenBioMed/blob/main/README-CN.md

主要就是将生物医药相关知识，文本、图像等多模态数据进行构建的行业大模型进行知识问答

简单在线网页体验

https://huggingface.co/spaces/Artples/BioMistral-7b-Chat

代码运行

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig

model_id = "BioMistral/BioMistral-7B"
bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_use_double_quant=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16
)

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    quantization_config=bnb_config,
    device_map="auto",
    trust_remote_code=True,
)

eval_tokenizer = AutoTokenizer.from_pretrained(model_id, add_bos_token=True, trust_remote_code=True)


eval_prompt = "The best way to "
model_input = eval_tokenizer(eval_prompt, return_tensors="pt").to("cuda")

model.eval()
with torch.no_grad():
    print(eval_tokenizer.decode(model.generate(**model_input, max_new_tokens=100, repetition_penalty=1.15)[0], skip_special_tokens=True))
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27