Compare commits
22
Commits
huggingface
...
patch-1
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
b20f43b2a8 | ||
|
|
ee7494c0b7 | ||
|
|
dcdc8351e7 | ||
|
|
a239abac50 | ||
|
|
1042d28e1f | ||
|
|
7b75422c26 | ||
|
|
99817e9040 | ||
|
|
c9fa26405d | ||
|
|
005232afa6 | ||
|
|
5b8cc5a899 | ||
|
|
a4137e7170 | ||
|
|
aaf44750d9 | ||
|
|
bd6eb90449 | ||
|
|
b5a48369a4 | ||
|
|
6b5bdbe98a | ||
|
|
69624c66d7 | ||
|
|
69be335d22 | ||
|
|
bb1e410cb4 | ||
|
|
9ccc53fa96 | ||
|
|
4b83486b3d | ||
|
|
417c8325de | ||
|
|
d51ae6abb2 |
@@ -0,0 +1,19 @@
|
||||
---
|
||||
name: Bug report
|
||||
about: Create a report to help us improve
|
||||
title: ''
|
||||
labels: ''
|
||||
assignees: ''
|
||||
|
||||
---
|
||||
|
||||
**Describe the bug 简述**
|
||||
|
||||
**Screen Shot 截图**
|
||||
|
||||
**Terminal Traceback 终端traceback(如果有)**
|
||||
|
||||
|
||||
Before submitting an issue 提交issue之前:
|
||||
- Please try to upgrade your code. 如果您的代码不是最新的,建议您先尝试更新代码
|
||||
- Please check project wiki for common problem solutions.项目[wiki](https://github.com/binary-husky/chatgpt_academic/wiki)有一些常见问题的解决方法
|
||||
@@ -0,0 +1,10 @@
|
||||
---
|
||||
name: Feature request
|
||||
about: Suggest an idea for this project
|
||||
title: ''
|
||||
labels: ''
|
||||
assignees: ''
|
||||
|
||||
---
|
||||
|
||||
|
||||
+2
-1
@@ -139,4 +139,5 @@ config_private.py
|
||||
gpt_log
|
||||
private.md
|
||||
private_upload
|
||||
other_llms
|
||||
other_llms
|
||||
cradle.py
|
||||
@@ -46,6 +46,8 @@ arxiv小助手 | [函数插件] 输入arxiv文章url即可一键翻译摘要+下
|
||||
图片显示 | 可以在markdown中显示图片
|
||||
多线程函数插件支持 | 支持多线调用chatgpt,一键处理海量文本或程序
|
||||
支持GPT输出的markdown表格 | 可以输出支持GPT的markdown表格
|
||||
启动暗色gradio[主题](https://github.com/binary-husky/chatgpt_academic/issues/173) | 在浏览器url后面添加```/?__dark-theme=true```可以切换dark主题
|
||||
huggingface免科学上网[在线体验](https://huggingface.co/spaces/qingxu98/gpt-academic) | 登陆huggingface后复制[此空间](https://huggingface.co/spaces/qingxu98/gpt-academic)
|
||||
…… | ……
|
||||
|
||||
</div>
|
||||
@@ -113,7 +115,7 @@ python -m pip install -r requirements.txt
|
||||
# (选择二.2)conda activate gptac_venv
|
||||
# (选择二.3)python -m pip install -r requirements.txt
|
||||
|
||||
# 备注:使用官方pip源或者阿里pip源,其他pip源(如清华pip)有可能出问题,临时换源方法:
|
||||
# 备注:使用官方pip源或者阿里pip源,其他pip源(如一些大学的pip)有可能出问题,临时换源方法:
|
||||
# python -m pip install -r requirements.txt -i https://mirrors.aliyun.com/pypi/simple/
|
||||
```
|
||||
|
||||
@@ -255,9 +257,24 @@ python check_proxy.py
|
||||
<img src="https://user-images.githubusercontent.com/96192199/227504931-19955f78-45cd-4d1c-adac-e71e50957915.png" height="400" >
|
||||
</div>
|
||||
|
||||
## Todo:
|
||||
|
||||
- (Top Priority) 调用另一个开源项目text-generation-webui的web接口,使用其他llm模型
|
||||
- 总结大工程源代码时,文本过长、token溢出的问题(目前的方法是直接二分丢弃处理溢出,过于粗暴,有效信息大量丢失)
|
||||
### 源代码转译英文
|
||||
|
||||
<div align="center">
|
||||
<img src="https://user-images.githubusercontent.com/96192199/229720562-fe6c3508-6142-4635-a83d-21eb3669baee.png" height="400" >
|
||||
</div>
|
||||
|
||||
## Todo 与 版本规划:
|
||||
|
||||
- version 3 (Todo):
|
||||
- - 支持gpt4和其他更多llm
|
||||
- version 2.3+ (Todo):
|
||||
- - 总结大工程源代码时文本过长、token溢出的问题
|
||||
- - 实现项目打包部署
|
||||
- - 函数插件参数接口优化
|
||||
- - 自更新
|
||||
- version 2.3: 增强多线程交互性
|
||||
- version 2.2: 函数插件支持热重载
|
||||
- version 2.1: 可折叠式布局
|
||||
- version 2.0: 引入模块化函数插件
|
||||
- version 1.0: 基础功能
|
||||
+114
-21
@@ -1,41 +1,126 @@
|
||||
import threading
|
||||
from predict import predict_no_ui_long_connection
|
||||
from toolbox import CatchException, write_results_to_file
|
||||
from toolbox import CatchException, write_results_to_file, report_execption
|
||||
|
||||
def extract_code_block_carefully(txt):
|
||||
splitted = txt.split('```')
|
||||
n_code_block_seg = len(splitted) - 1
|
||||
if n_code_block_seg <= 1: return txt
|
||||
# 剩下的情况都开头除去 ``` 结尾除去一次 ```
|
||||
txt_out = '```'.join(splitted[1:-1])
|
||||
return txt_out
|
||||
|
||||
def breakdown_txt_to_satisfy_token_limit(txt, limit, must_break_at_empty_line=True):
|
||||
from transformers import GPT2TokenizerFast
|
||||
tokenizer = GPT2TokenizerFast.from_pretrained("gpt2")
|
||||
get_token_cnt = lambda txt: len(tokenizer(txt)["input_ids"])
|
||||
def cut(txt_tocut, must_break_at_empty_line): # 递归
|
||||
if get_token_cnt(txt_tocut) <= limit:
|
||||
return [txt_tocut]
|
||||
else:
|
||||
lines = txt_tocut.split('\n')
|
||||
estimated_line_cut = limit / get_token_cnt(txt_tocut) * len(lines)
|
||||
estimated_line_cut = int(estimated_line_cut)
|
||||
for cnt in reversed(range(estimated_line_cut)):
|
||||
if must_break_at_empty_line:
|
||||
if lines[cnt] != "": continue
|
||||
print(cnt)
|
||||
prev = "\n".join(lines[:cnt])
|
||||
post = "\n".join(lines[cnt:])
|
||||
if get_token_cnt(prev) < limit: break
|
||||
if cnt == 0:
|
||||
print('what the f?')
|
||||
raise RuntimeError("存在一行极长的文本!")
|
||||
print(len(post))
|
||||
# 列表递归接龙
|
||||
result = [prev]
|
||||
result.extend(cut(post, must_break_at_empty_line))
|
||||
return result
|
||||
try:
|
||||
return cut(txt, must_break_at_empty_line=True)
|
||||
except RuntimeError:
|
||||
return cut(txt, must_break_at_empty_line=False)
|
||||
|
||||
|
||||
def break_txt_into_half_at_some_linebreak(txt):
|
||||
lines = txt.split('\n')
|
||||
n_lines = len(lines)
|
||||
pre = lines[:(n_lines//2)]
|
||||
post = lines[(n_lines//2):]
|
||||
return "\n".join(pre), "\n".join(post)
|
||||
|
||||
|
||||
@CatchException
|
||||
def 全项目切换英文(txt, top_p, temperature, chatbot, history, sys_prompt, WEB_PORT):
|
||||
history = [] # 清空历史,以免输入溢出
|
||||
# 集合文件
|
||||
import time, glob, os
|
||||
# 第1步:清空历史,以免输入溢出
|
||||
history = []
|
||||
|
||||
# 第2步:尝试导入依赖,如果缺少依赖,则给出安装建议
|
||||
try:
|
||||
import openai, transformers
|
||||
except:
|
||||
report_execption(chatbot, history,
|
||||
a = f"解析项目: {txt}",
|
||||
b = f"导入软件依赖失败。使用该模块需要额外依赖,安装方法```pip install --upgrade openai transformers```。")
|
||||
yield chatbot, history, '正常'
|
||||
return
|
||||
|
||||
# 第3步:集合文件
|
||||
import time, glob, os, shutil, re, openai
|
||||
os.makedirs('gpt_log/generated_english_version', exist_ok=True)
|
||||
os.makedirs('gpt_log/generated_english_version/crazy_functions', exist_ok=True)
|
||||
file_manifest = [f for f in glob.glob('./*.py') if ('test_project' not in f) and ('gpt_log' not in f)] + \
|
||||
[f for f in glob.glob('./crazy_functions/*.py') if ('test_project' not in f) and ('gpt_log' not in f)]
|
||||
# file_manifest = ['./toolbox.py']
|
||||
i_say_show_user_buffer = []
|
||||
|
||||
# 随便显示点什么防止卡顿的感觉
|
||||
# 第4步:随便显示点什么防止卡顿的感觉
|
||||
for index, fp in enumerate(file_manifest):
|
||||
# if 'test_project' in fp: continue
|
||||
with open(fp, 'r', encoding='utf-8') as f:
|
||||
file_content = f.read()
|
||||
i_say_show_user =f'[{index}/{len(file_manifest)}] 接下来请将以下代码中包含的所有中文转化为英文,只输出代码: {os.path.abspath(fp)}'
|
||||
i_say_show_user =f'[{index}/{len(file_manifest)}] 接下来请将以下代码中包含的所有中文转化为英文,只输出转化后的英文代码,请用代码块输出代码: {os.path.abspath(fp)}'
|
||||
i_say_show_user_buffer.append(i_say_show_user)
|
||||
chatbot.append((i_say_show_user, "[Local Message] 等待多线程操作,中间过程不予显示."))
|
||||
yield chatbot, history, '正常'
|
||||
|
||||
# 任务函数
|
||||
|
||||
# 第5步:Token限制下的截断与处理
|
||||
MAX_TOKEN = 2500
|
||||
# from transformers import GPT2TokenizerFast
|
||||
# print('加载tokenizer中')
|
||||
# tokenizer = GPT2TokenizerFast.from_pretrained("gpt2")
|
||||
# get_token_cnt = lambda txt: len(tokenizer(txt)["input_ids"])
|
||||
# print('加载tokenizer结束')
|
||||
|
||||
|
||||
# 第6步:任务函数
|
||||
mutable_return = [None for _ in file_manifest]
|
||||
observe_window = [[""] for _ in file_manifest]
|
||||
def thread_worker(fp,index):
|
||||
if index > 10:
|
||||
time.sleep(60)
|
||||
print('Openai 限制免费用户每分钟20次请求,降低请求频率中。')
|
||||
with open(fp, 'r', encoding='utf-8') as f:
|
||||
file_content = f.read()
|
||||
i_say = f'接下来请将以下代码中包含的所有中文转化为英文,只输出代码,文件名是{fp},文件代码是 ```{file_content}```'
|
||||
# ** gpt request **
|
||||
gpt_say = predict_no_ui_long_connection(inputs=i_say, top_p=top_p, temperature=temperature, history=history, sys_prompt=sys_prompt)
|
||||
mutable_return[index] = gpt_say
|
||||
i_say_template = lambda fp, file_content: f'接下来请将以下代码中包含的所有中文转化为英文,只输出代码,文件名是{fp},文件代码是 ```{file_content}```'
|
||||
try:
|
||||
gpt_say = ""
|
||||
# 分解代码文件
|
||||
file_content_breakdown = breakdown_txt_to_satisfy_token_limit(file_content, MAX_TOKEN)
|
||||
for file_content_partial in file_content_breakdown:
|
||||
i_say = i_say_template(fp, file_content_partial)
|
||||
# # ** gpt request **
|
||||
gpt_say_partial = predict_no_ui_long_connection(inputs=i_say, top_p=top_p, temperature=temperature, history=[], sys_prompt=sys_prompt, observe_window=observe_window[index])
|
||||
gpt_say_partial = extract_code_block_carefully(gpt_say_partial)
|
||||
gpt_say += gpt_say_partial
|
||||
mutable_return[index] = gpt_say
|
||||
except ConnectionAbortedError as token_exceed_err:
|
||||
print('至少一个线程任务Token溢出而失败', e)
|
||||
except Exception as e:
|
||||
print('至少一个线程任务意外失败', e)
|
||||
|
||||
# 所有线程同时开始执行任务函数
|
||||
# 第7步:所有线程同时开始执行任务函数
|
||||
handles = [threading.Thread(target=thread_worker, args=(fp,index)) for index, fp in enumerate(file_manifest)]
|
||||
for h in handles:
|
||||
h.daemon = True
|
||||
@@ -43,19 +128,23 @@ def 全项目切换英文(txt, top_p, temperature, chatbot, history, sys_prompt,
|
||||
chatbot.append(('开始了吗?', f'多线程操作已经开始'))
|
||||
yield chatbot, history, '正常'
|
||||
|
||||
# 循环轮询各个线程是否执行完毕
|
||||
# 第8步:循环轮询各个线程是否执行完毕
|
||||
cnt = 0
|
||||
while True:
|
||||
time.sleep(1)
|
||||
cnt += 1
|
||||
time.sleep(0.2)
|
||||
th_alive = [h.is_alive() for h in handles]
|
||||
if not any(th_alive): break
|
||||
stat = ['执行中' if alive else '已完成' for alive in th_alive]
|
||||
stat_str = '|'.join(stat)
|
||||
cnt += 1
|
||||
chatbot[-1] = (chatbot[-1][0], f'多线程操作已经开始,完成情况: {stat_str}' + ''.join(['.']*(cnt%4)))
|
||||
# 更好的UI视觉效果
|
||||
observe_win = []
|
||||
for thread_index, alive in enumerate(th_alive):
|
||||
observe_win.append("[ ..."+observe_window[thread_index][0][-60:].replace('\n','').replace('```','...').replace(' ','.').replace('<br/>','.....').replace('$','.')+"... ]")
|
||||
stat = [f'执行中: {obs}\n\n' if alive else '已完成\n\n' for alive, obs in zip(th_alive, observe_win)]
|
||||
stat_str = ''.join(stat)
|
||||
chatbot[-1] = (chatbot[-1][0], f'多线程操作已经开始,完成情况: \n\n{stat_str}' + ''.join(['.']*(cnt%10+1)))
|
||||
yield chatbot, history, '正常'
|
||||
|
||||
# 把结果写入文件
|
||||
# 第9步:把结果写入文件
|
||||
for index, h in enumerate(handles):
|
||||
h.join() # 这里其实不需要join了,肯定已经都结束了
|
||||
fp = file_manifest[index]
|
||||
@@ -63,13 +152,17 @@ def 全项目切换英文(txt, top_p, temperature, chatbot, history, sys_prompt,
|
||||
i_say_show_user = i_say_show_user_buffer[index]
|
||||
|
||||
where_to_relocate = f'gpt_log/generated_english_version/{fp}'
|
||||
with open(where_to_relocate, 'w+', encoding='utf-8') as f: f.write(gpt_say.lstrip('```').rstrip('```'))
|
||||
if gpt_say is not None:
|
||||
with open(where_to_relocate, 'w+', encoding='utf-8') as f:
|
||||
f.write(gpt_say)
|
||||
else: # 失败
|
||||
shutil.copyfile(file_manifest[index], where_to_relocate)
|
||||
chatbot.append((i_say_show_user, f'[Local Message] 已完成{os.path.abspath(fp)}的转化,\n\n存入{os.path.abspath(where_to_relocate)}'))
|
||||
history.append(i_say_show_user); history.append(gpt_say)
|
||||
yield chatbot, history, '正常'
|
||||
time.sleep(1)
|
||||
|
||||
# 备份一个文件
|
||||
# 第10步:备份一个文件
|
||||
res = write_results_to_file(history)
|
||||
chatbot.append(("生成一份任务执行报告", res))
|
||||
yield chatbot, history, '正常'
|
||||
|
||||
@@ -119,8 +119,8 @@ def 解析一个C项目的头文件(txt, top_p, temperature, chatbot, history, s
|
||||
report_execption(chatbot, history, a = f"解析项目: {txt}", b = f"找不到本地项目或无权访问: {txt}")
|
||||
yield chatbot, history, '正常'
|
||||
return
|
||||
file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.h', recursive=True)] # + \
|
||||
# [f for f in glob.glob(f'{project_folder}/**/*.cpp', recursive=True)] + \
|
||||
file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.h', recursive=True)] + \
|
||||
[f for f in glob.glob(f'{project_folder}/**/*.hpp', recursive=True)] #+ \
|
||||
# [f for f in glob.glob(f'{project_folder}/**/*.c', recursive=True)]
|
||||
if len(file_manifest) == 0:
|
||||
report_execption(chatbot, history, a = f"解析项目: {txt}", b = f"找不到任何.h头文件: {txt}")
|
||||
@@ -141,6 +141,7 @@ def 解析一个C项目(txt, top_p, temperature, chatbot, history, systemPromptT
|
||||
return
|
||||
file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.h', recursive=True)] + \
|
||||
[f for f in glob.glob(f'{project_folder}/**/*.cpp', recursive=True)] + \
|
||||
[f for f in glob.glob(f'{project_folder}/**/*.hpp', recursive=True)] + \
|
||||
[f for f in glob.glob(f'{project_folder}/**/*.c', recursive=True)]
|
||||
if len(file_manifest) == 0:
|
||||
report_execption(chatbot, history, a = f"解析项目: {txt}", b = f"找不到任何.h头文件: {txt}")
|
||||
|
||||
+1
-1
@@ -43,7 +43,7 @@ def get_crazy_functionals():
|
||||
"AsButton": False, # 加入下拉菜单中
|
||||
"Function": 解析一个Java项目
|
||||
},
|
||||
"解析整个Java项目": {
|
||||
"解析整个React项目": {
|
||||
"Color": "stop", # 按钮颜色
|
||||
"AsButton": False, # 加入下拉菜单中
|
||||
"Function": 解析一个Rect项目
|
||||
|
||||
@@ -134,12 +134,14 @@ with gr.Blocks(theme=set_theme, analytics_enabled=False, css=advanced_css) as de
|
||||
# gradio的inbrowser触发不太稳定,回滚代码到原始的浏览器打开函数
|
||||
def auto_opentab_delay():
|
||||
import threading, webbrowser, time
|
||||
print(f"如果浏览器没有自动打开,请复制并转到以下URL: http://localhost:{PORT}")
|
||||
print(f"如果浏览器没有自动打开,请复制并转到以下URL:")
|
||||
print(f"\t(亮色主体): http://localhost:{PORT}")
|
||||
print(f"\t(暗色主体): http://localhost:{PORT}/?__dark-theme=true")
|
||||
def open():
|
||||
time.sleep(2)
|
||||
webbrowser.open_new_tab(f"http://localhost:{PORT}")
|
||||
webbrowser.open_new_tab(f"http://localhost:{PORT}/?__dark-theme=true")
|
||||
threading.Thread(target=open, name="open-browser", daemon=True).start()
|
||||
|
||||
auto_opentab_delay()
|
||||
demo.title = "ChatGPT 学术优化"
|
||||
demo.queue(concurrency_count=CONCURRENT_COUNT).launch(server_name="0.0.0.0", share=True, server_port=PORT, auth=AUTHENTICATION)
|
||||
demo.queue(concurrency_count=CONCURRENT_COUNT).launch(server_name="0.0.0.0", share=Flase, server_port=8080, auth=chunzhi233233)
|
||||
|
||||
+6
-2
@@ -71,9 +71,10 @@ def predict_no_ui(inputs, top_p, temperature, history=[], sys_prompt=""):
|
||||
raise ConnectionAbortedError("Json解析不合常规,可能是文本过长" + response.text)
|
||||
|
||||
|
||||
def predict_no_ui_long_connection(inputs, top_p, temperature, history=[], sys_prompt=""):
|
||||
def predict_no_ui_long_connection(inputs, top_p, temperature, history=[], sys_prompt="", observe_window=None):
|
||||
"""
|
||||
发送至chatGPT,等待回复,一次性完成,不显示中间过程。但内部用stream的方法避免有人中途掐网线。
|
||||
observe_window:用于负责跨越线程传递已经输出的部分,大部分时候仅仅为了fancy的视觉效果,留空即可
|
||||
"""
|
||||
headers, payload = generate_payload(inputs, top_p, temperature, history, system_prompt=sys_prompt, stream=True)
|
||||
|
||||
@@ -105,7 +106,10 @@ def predict_no_ui_long_connection(inputs, top_p, temperature, history=[], sys_pr
|
||||
delta = json_data["delta"]
|
||||
if len(delta) == 0: break
|
||||
if "role" in delta: continue
|
||||
if "content" in delta: result += delta["content"]; print(delta["content"], end='')
|
||||
if "content" in delta:
|
||||
result += delta["content"]
|
||||
print(delta["content"], end='')
|
||||
if observe_window is not None: observe_window[0] += delta["content"]
|
||||
else: raise RuntimeError("意外Json结构:"+delta)
|
||||
if json_data['finish_reason'] == 'length':
|
||||
raise ConnectionAbortedError("正常结束,但显示Token不足。")
|
||||
|
||||
@@ -3,3 +3,5 @@ requests[socks]
|
||||
mdtex2html
|
||||
Markdown
|
||||
latex2mathml
|
||||
openai
|
||||
transformers
|
||||
|
||||
+1
-1
@@ -303,7 +303,7 @@ def on_file_uploaded(files, chatbot, txt):
|
||||
def on_report_generated(files, chatbot):
|
||||
from toolbox import find_recent_files
|
||||
report_files = find_recent_files('gpt_log')
|
||||
if len(report_files) == 0: return files, chatbot
|
||||
if len(report_files) == 0: return None, chatbot
|
||||
# files.extend(report_files)
|
||||
chatbot.append(['汇总报告如何远程获取?', '汇总报告已经添加到右侧“文件上传区”(可能处于折叠状态),请查收。'])
|
||||
return report_files, chatbot
|
||||
|
||||
Reference in New Issue
Block a user