27 lines
1.2 KiB
Python
27 lines
1.2 KiB
Python
import torch
|
||
from transformers import AutoTokenizer, AutoModelForCausalLM
|
||
from openxlab.model import download
|
||
|
||
download(model_repo='jujimeizuo/EmoLLM_Model',
|
||
output='model')
|
||
|
||
model_name_or_path = "model"
|
||
|
||
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, trust_remote_code=True)
|
||
model = AutoModelForCausalLM.from_pretrained(model_name_or_path, trust_remote_code=True, torch_dtype=torch.bfloat16, device_map='auto')
|
||
model = model.eval()
|
||
|
||
system_prompt = "你是一个由aJupyter、Farewell、jujimeizuo、Smiling&Weeping研发（排名按字母顺序排序，不分先后）、散步提供技术支持、上海人工智能实验室提供支持开发的心理健康大模型。现在你是一个心理专家，我有一些心理问题，请你用专业的知识帮我解决。"
|
||
|
||
messages = [(system_prompt, '')]
|
||
|
||
print("=============Welcome to InternLM chatbot, type 'exit' to exit.=============")
|
||
|
||
while True:
|
||
input_text = input("User >>> ")
|
||
input_text.replace(' ', '')
|
||
if input_text == "exit":
|
||
break
|
||
response, history = model.chat(tokenizer, input_text, history=messages)
|
||
messages.append((input_text, response))
|
||
print(f"robot >>> {response}") |