Add application file
Browse files
chat.py
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# -*- coding: utf-8 -*-
|
2 |
+
"""
|
3 |
+
@Time : 2023/3/13 16:18
|
4 |
+
@Auth : mxzh
|
5 |
+
@DESC:使用gradio chatbot组件部署 openai接口, 流式返回
|
6 |
+
"""
|
7 |
+
|
8 |
+
import gradio as gr
|
9 |
+
import torch
|
10 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
11 |
+
from transformers.generation.utils import GenerationConfig
|
12 |
+
|
13 |
+
|
14 |
+
def load_model():
|
15 |
+
tokenizer = AutoTokenizer.from_pretrained("baichuan-inc/Baichuan2-7B-Chat", use_fast=False, trust_remote_code=True)
|
16 |
+
model = AutoModelForCausalLM.from_pretrained("baichuan-inc/Baichuan2-7B-Chat", device_map="auto", torch_dtype=torch.bfloat16, trust_remote_code=True)
|
17 |
+
model.generation_config = GenerationConfig.from_pretrained("baichuan-inc/Baichuan2-7B-Chat")
|
18 |
+
return model, tokenizer
|
19 |
+
|
20 |
+
|
21 |
+
def chat(content):
|
22 |
+
messages = []
|
23 |
+
messages.append({"role": "user", "content": content})
|
24 |
+
response = model.chat(tokenizer, messages)
|
25 |
+
return response
|
26 |
+
|
27 |
+
|
28 |
+
if __name__ == "__main__":
|
29 |
+
model, tokenizer = load_model()
|
30 |
+
iface = gr.Interface(fn=chat, inputs="text", outputs="text")
|
31 |
+
iface.launch()
|
32 |
+
|