Browse Source

Merge pull request #954 from gangqing/gangqing-patch-1

修复预测结果为空的bug
dev
Zhengxiao Du 2 years ago committed by GitHub
parent
commit
79aff8dff4
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
  1. 1
      PROJECT.md
  2. 1
      cli_demo.py
  3. 6
      ptuning/main.py
  4. 4
      web_demo2.py

1
PROJECT.md

@ -17,6 +17,7 @@
* [ChatGLM-web](https://github.com/NCZkevin/chatglm-web):基于FastAPI和Vue3搭建的ChatGLM演示网站(支持chatglm流式输出、前端调整模型参数、上下文选择、保存图片、知识库问答等功能)
* [ChatGLM-6B-Engineering](https://github.com/LemonQu-GIT/ChatGLM-6B-Engineering):基于 ChatGLM-6B 后期调教,网络爬虫及 [Stable Diffusion](https://github.com/AUTOMATIC1111/stable-diffusion-webui) 实现的网络搜索及图片生成
* [ChatGLM-OpenAI-API](https://github.com/ninehills/chatglm-openai-api): 将 ChatGLM-6B 封装为 OpenAI API 风格,并通过 ngrok/cloudflare 对外提供服务,从而将 ChatGLM 快速集成到 OpenAI 的各种生态中。
* [ChatSQL](https://github.com/yysirs/ChatSQL): 基于ChatGLM+SBERT实现NL2SQL本地化,并直接连接数据库查询数据返回结果,使得生成的SQL语句更具有实用性。
对 ChatGLM-6B 进行微调的开源项目:
* [InstructGLM](https://github.com/yanqiangmiffy/InstructGLM):基于ChatGLM-6B进行指令学习,汇总开源中英文指令数据,基于Lora进行指令数据微调,开放了Alpaca、Belle微调后的Lora权重,修复web_demo重复问题

1
cli_demo.py

@ -2,6 +2,7 @@ import os
import platform
import signal
from transformers import AutoTokenizer, AutoModel
import readline
tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True)
model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).half().cuda()

6
ptuning/main.py

@ -382,9 +382,10 @@ def main():
# Evaluation
results = {}
max_seq_length = data_args.max_source_length + data_args.max_target_length + 1
if training_args.do_eval:
logger.info("*** Evaluate ***")
metrics = trainer.evaluate(metric_key_prefix="eval", do_sample=True, top_p=0.7, max_length=512, temperature=0.95)
metrics = trainer.evaluate(metric_key_prefix="eval", do_sample=True, top_p=0.7, max_length=max_seq_length, temperature=0.95)
max_eval_samples = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(eval_dataset)
metrics["eval_samples"] = min(max_eval_samples, len(eval_dataset))
@ -393,8 +394,7 @@ def main():
if training_args.do_predict:
logger.info("*** Predict ***")
predict_results = trainer.predict(predict_dataset, metric_key_prefix="predict", max_length=512, do_sample=True, top_p=0.7, temperature=0.95)
predict_results = trainer.predict(predict_dataset, metric_key_prefix="predict", max_length=max_seq_length, do_sample=True, top_p=0.7, temperature=0.95)
metrics = predict_results.metrics
max_predict_samples = (
data_args.max_predict_samples if data_args.max_predict_samples is not None else len(predict_dataset)

4
web_demo2.py

@ -28,6 +28,8 @@ def predict(input, max_length, top_p, temperature, history=None):
with container:
if len(history) > 0:
if len(history)>MAX_BOXES:
history = history[-MAX_TURNS:]
for i, (query, response) in enumerate(history):
message(query, avatar_style="big-smile", key=str(i) + "_user")
message(response, avatar_style="bottts", key=str(i))
@ -66,4 +68,4 @@ if 'state' not in st.session_state:
if st.button("发送", key="predict"):
with st.spinner("AI正在思考,请稍等........"):
# text generation
st.session_state["state"] = predict(prompt_text, max_length, top_p, temperature, st.session_state["state"])
st.session_state["state"] = predict(prompt_text, max_length, top_p, temperature, st.session_state["state"])

Loading…
Cancel
Save