Spaces:
Paused
Paused
Merge branch 'test-3-26'
Browse files- Dockerfile +1 -5
- README.md +23 -7
- crazy_functions/解析项目源代码.py +6 -6
- crazy_functions/读文章写摘要.py +3 -3
- main.py +3 -1
- predict.py +25 -1
- toolbox.py +6 -1
Dockerfile
CHANGED
|
@@ -1,8 +1,4 @@
|
|
| 1 |
-
FROM
|
| 2 |
-
|
| 3 |
-
RUN apt-get update && \
|
| 4 |
-
apt-get install -y python3 python3-pip && \
|
| 5 |
-
rm -rf /var/lib/apt/lists/*
|
| 6 |
|
| 7 |
RUN echo '[global]' > /etc/pip.conf && \
|
| 8 |
echo 'index-url = https://mirrors.aliyun.com/pypi/simple/' >> /etc/pip.conf && \
|
|
|
|
| 1 |
+
FROM python:3.11
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
|
| 3 |
RUN echo '[global]' > /etc/pip.conf && \
|
| 4 |
echo 'index-url = https://mirrors.aliyun.com/pypi/simple/' >> /etc/pip.conf && \
|
README.md
CHANGED
|
@@ -60,7 +60,7 @@ chat分析报告生成 | [实验性功能] 运行后自动生成总结汇报
|
|
| 60 |
|
| 61 |
## 直接运行 (Windows or Linux or MacOS)
|
| 62 |
|
| 63 |
-
```
|
| 64 |
# 下载项目
|
| 65 |
git clone https://github.com/binary-husky/chatgpt_academic.git
|
| 66 |
cd chatgpt_academic
|
|
@@ -73,9 +73,16 @@ python -m pip install -r requirements.txt
|
|
| 73 |
python main.py
|
| 74 |
|
| 75 |
# 测试实验性功能
|
| 76 |
-
|
| 77 |
-
input区域 输入 ./crazy_functions/test_project/
|
| 78 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 79 |
```
|
| 80 |
|
| 81 |
|
|
@@ -93,9 +100,18 @@ docker build -t gpt-academic .
|
|
| 93 |
docker run --rm -it --net=host gpt-academic
|
| 94 |
|
| 95 |
# 测试实验性功能
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 99 |
```
|
| 100 |
|
| 101 |
|
|
|
|
| 60 |
|
| 61 |
## 直接运行 (Windows or Linux or MacOS)
|
| 62 |
|
| 63 |
+
``` sh
|
| 64 |
# 下载项目
|
| 65 |
git clone https://github.com/binary-husky/chatgpt_academic.git
|
| 66 |
cd chatgpt_academic
|
|
|
|
| 73 |
python main.py
|
| 74 |
|
| 75 |
# 测试实验性功能
|
| 76 |
+
## 测试C++项目头文件分析
|
| 77 |
+
input区域 输入 ./crazy_functions/test_project/cpp/libJPG , 然后点击 "[实验] 解析整个C++项目(input输入项目根路径)"
|
| 78 |
+
## 测试给Latex项目写摘要
|
| 79 |
+
input区域 输入 ./crazy_functions/test_project/latex/attention , 然后点击 "[实验] 读tex论文写摘要(input输入项目根路径)"
|
| 80 |
+
## 测试Python项目分析
|
| 81 |
+
input区域 输入 ./crazy_functions/test_project/python/dqn , 然后点击 "[实验] 解析整个py项目(input输入项目根路径)"
|
| 82 |
+
## 测试自我代码解读
|
| 83 |
+
点击 "[实验] 请解析并解构此项目本身"
|
| 84 |
+
## 测试实验功能模板函数(要求gpt回答几个数的平方是什么),您可以根据此函数为模板,实现更复杂的功能
|
| 85 |
+
点击 "[实验] 实验功能函数模板"
|
| 86 |
```
|
| 87 |
|
| 88 |
|
|
|
|
| 100 |
docker run --rm -it --net=host gpt-academic
|
| 101 |
|
| 102 |
# 测试实验性功能
|
| 103 |
+
## 测试自我代码解读
|
| 104 |
+
点击 "[实验] 请解析并解构此项目本身"
|
| 105 |
+
## 测试实验功能模板函数(要求gpt回答几个数的平方是什么),您可以根据此函数为模板,实现更复杂的功能
|
| 106 |
+
点击 "[实验] 实验功能函数模板"
|
| 107 |
+
##(请注意在docker中运行时,需要额外注意程序的文件访问权限问题)
|
| 108 |
+
## 测试C++项目头文件分析
|
| 109 |
+
input区域 输入 ./crazy_functions/test_project/cpp/libJPG , 然后点击 "[实验] 解析整个C++项目(input输入项目根路径)"
|
| 110 |
+
## 测试给Latex项目写摘要
|
| 111 |
+
input区域 输入 ./crazy_functions/test_project/latex/attention , 然后点击 "[实验] 读tex论文写摘要(input输入项目根路径)"
|
| 112 |
+
## 测试Python项目分析
|
| 113 |
+
input区域 输入 ./crazy_functions/test_project/python/dqn , 然后点击 "[实验] 解析整个py项目(input输入项目根路径)"
|
| 114 |
+
|
| 115 |
```
|
| 116 |
|
| 117 |
|
crazy_functions/解析项目源代码.py
CHANGED
|
@@ -9,9 +9,9 @@ def 解析源代码(file_manifest, project_folder, top_p, temperature, chatbot,
|
|
| 9 |
with open(fp, 'r', encoding='utf-8') as f:
|
| 10 |
file_content = f.read()
|
| 11 |
|
| 12 |
-
|
| 13 |
-
i_say =
|
| 14 |
-
i_say_show_user =
|
| 15 |
chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
|
| 16 |
yield chatbot, history, '正常'
|
| 17 |
|
|
@@ -56,9 +56,9 @@ def 解析项目本身(txt, top_p, temperature, chatbot, history, systemPromptTx
|
|
| 56 |
with open(fp, 'r', encoding='utf-8') as f:
|
| 57 |
file_content = f.read()
|
| 58 |
|
| 59 |
-
|
| 60 |
-
i_say =
|
| 61 |
-
i_say_show_user =
|
| 62 |
chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
|
| 63 |
yield chatbot, history, '正常'
|
| 64 |
|
|
|
|
| 9 |
with open(fp, 'r', encoding='utf-8') as f:
|
| 10 |
file_content = f.read()
|
| 11 |
|
| 12 |
+
prefix = "接下来请你逐文件分析下面的工程" if index==0 else ""
|
| 13 |
+
i_say = prefix + f'请对下面的程序文件做一个概述文件名是{os.path.relpath(fp, project_folder)},文件代码是 ```{file_content}```'
|
| 14 |
+
i_say_show_user = prefix + f'[{index}/{len(file_manifest)}] 请对下面的程序文件做一个概述: {os.path.abspath(fp)}'
|
| 15 |
chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
|
| 16 |
yield chatbot, history, '正常'
|
| 17 |
|
|
|
|
| 56 |
with open(fp, 'r', encoding='utf-8') as f:
|
| 57 |
file_content = f.read()
|
| 58 |
|
| 59 |
+
prefix = "接下来请你分析自己的程序构成,别紧张," if index==0 else ""
|
| 60 |
+
i_say = prefix + f'请对下面的程序文件做一个概述文件名是{fp},文件代码是 ```{file_content}```'
|
| 61 |
+
i_say_show_user = prefix + f'[{index}/{len(file_manifest)}] 请对下面的程序文件做一个概述: {os.path.abspath(fp)}'
|
| 62 |
chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
|
| 63 |
yield chatbot, history, '正常'
|
| 64 |
|
crazy_functions/读文章写摘要.py
CHANGED
|
@@ -10,9 +10,9 @@ def 解析Paper(file_manifest, project_folder, top_p, temperature, chatbot, hist
|
|
| 10 |
with open(fp, 'r', encoding='utf-8') as f:
|
| 11 |
file_content = f.read()
|
| 12 |
|
| 13 |
-
|
| 14 |
-
i_say =
|
| 15 |
-
i_say_show_user =
|
| 16 |
chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
|
| 17 |
print('[1] yield chatbot, history')
|
| 18 |
yield chatbot, history, '正常'
|
|
|
|
| 10 |
with open(fp, 'r', encoding='utf-8') as f:
|
| 11 |
file_content = f.read()
|
| 12 |
|
| 13 |
+
prefix = "接下来请你逐文件分析下面的论文文件,概括其内容" if index==0 else ""
|
| 14 |
+
i_say = prefix + f'请对下面的文章片段用中文做一个概述,文件名是{os.path.relpath(fp, project_folder)},文章内容是 ```{file_content}```'
|
| 15 |
+
i_say_show_user = prefix + f'[{index}/{len(file_manifest)}] 请对下面的文章片段做一个概述: {os.path.abspath(fp)}'
|
| 16 |
chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
|
| 17 |
print('[1] yield chatbot, history')
|
| 18 |
yield chatbot, history, '正常'
|
main.py
CHANGED
|
@@ -13,9 +13,11 @@ PORT = find_free_port() if WEB_PORT <= 0 else WEB_PORT
|
|
| 13 |
initial_prompt = "Serve me as a writing and programming assistant."
|
| 14 |
title_html = """<h1 align="center">ChatGPT 学术优化</h1>"""
|
| 15 |
|
|
|
|
| 16 |
import logging
|
| 17 |
os.makedirs('gpt_log', exist_ok=True)
|
| 18 |
-
logging.basicConfig(filename='gpt_log/chat_secrets.log', level=logging.INFO, encoding='utf-8')
|
|
|
|
| 19 |
print('所有问询记录将自动保存在本地目录./gpt_log/chat_secrets.log,请注意自我隐私保护哦!')
|
| 20 |
|
| 21 |
# 一些普通功能模块
|
|
|
|
| 13 |
initial_prompt = "Serve me as a writing and programming assistant."
|
| 14 |
title_html = """<h1 align="center">ChatGPT 学术优化</h1>"""
|
| 15 |
|
| 16 |
+
# 问询记录,python 版本建议3.9+(越新越好)
|
| 17 |
import logging
|
| 18 |
os.makedirs('gpt_log', exist_ok=True)
|
| 19 |
+
try:logging.basicConfig(filename='gpt_log/chat_secrets.log', level=logging.INFO, encoding='utf-8')
|
| 20 |
+
except:logging.basicConfig(filename='gpt_log/chat_secrets.log', level=logging.INFO)
|
| 21 |
print('所有问询记录将自动保存在本地目录./gpt_log/chat_secrets.log,请注意自我隐私保护哦!')
|
| 22 |
|
| 23 |
# 一些普通功能模块
|
predict.py
CHANGED
|
@@ -15,6 +15,9 @@ except: from config import proxies, API_URL, API_KEY, TIMEOUT_SECONDS, MAX_RETRY
|
|
| 15 |
timeout_bot_msg = '[local] Request timeout, network error. please check proxy settings in config.py.'
|
| 16 |
|
| 17 |
def get_full_error(chunk, stream_response):
|
|
|
|
|
|
|
|
|
|
| 18 |
while True:
|
| 19 |
try:
|
| 20 |
chunk += next(stream_response)
|
|
@@ -23,6 +26,16 @@ def get_full_error(chunk, stream_response):
|
|
| 23 |
return chunk
|
| 24 |
|
| 25 |
def predict_no_ui(inputs, top_p, temperature, history=[]):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
headers, payload = generate_payload(inputs, top_p, temperature, history, system_prompt="", stream=False)
|
| 27 |
|
| 28 |
retry = 0
|
|
@@ -47,7 +60,15 @@ def predict_no_ui(inputs, top_p, temperature, history=[]):
|
|
| 47 |
|
| 48 |
def predict(inputs, top_p, temperature, chatbot=[], history=[], system_prompt='',
|
| 49 |
stream = True, additional_fn=None):
|
| 50 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
if additional_fn is not None:
|
| 52 |
import functional
|
| 53 |
importlib.reload(functional)
|
|
@@ -115,6 +136,9 @@ def predict(inputs, top_p, temperature, chatbot=[], history=[], system_prompt=''
|
|
| 115 |
return
|
| 116 |
|
| 117 |
def generate_payload(inputs, top_p, temperature, history, system_prompt, stream):
|
|
|
|
|
|
|
|
|
|
| 118 |
headers = {
|
| 119 |
"Content-Type": "application/json",
|
| 120 |
"Authorization": f"Bearer {API_KEY}"
|
|
|
|
| 15 |
timeout_bot_msg = '[local] Request timeout, network error. please check proxy settings in config.py.'
|
| 16 |
|
| 17 |
def get_full_error(chunk, stream_response):
|
| 18 |
+
"""
|
| 19 |
+
获取完整的从Openai返回的报错
|
| 20 |
+
"""
|
| 21 |
while True:
|
| 22 |
try:
|
| 23 |
chunk += next(stream_response)
|
|
|
|
| 26 |
return chunk
|
| 27 |
|
| 28 |
def predict_no_ui(inputs, top_p, temperature, history=[]):
|
| 29 |
+
"""
|
| 30 |
+
发送至chatGPT,等待回复,一次性完成,不显示中间过程。
|
| 31 |
+
predict函数的简化版。
|
| 32 |
+
用于payload比较大的情况,或者用于实现多线、带嵌套的复杂功能。
|
| 33 |
+
|
| 34 |
+
inputs 是本次问询的输入
|
| 35 |
+
top_p, temperature是chatGPT的内部调优参数
|
| 36 |
+
history 是之前的对话列表
|
| 37 |
+
(注意无论是inputs还是history,内容太长了都会触发token数量溢出的错误,然后raise ConnectionAbortedError)
|
| 38 |
+
"""
|
| 39 |
headers, payload = generate_payload(inputs, top_p, temperature, history, system_prompt="", stream=False)
|
| 40 |
|
| 41 |
retry = 0
|
|
|
|
| 60 |
|
| 61 |
def predict(inputs, top_p, temperature, chatbot=[], history=[], system_prompt='',
|
| 62 |
stream = True, additional_fn=None):
|
| 63 |
+
"""
|
| 64 |
+
发送至chatGPT,流式获取输出。
|
| 65 |
+
用于基础的对话功能。
|
| 66 |
+
inputs 是本次问询的输入
|
| 67 |
+
top_p, temperature是chatGPT的内部调优参数
|
| 68 |
+
history 是之前的对话列表(注意无论是inputs还是history,内容太长了都会触发token数量溢出的错误)
|
| 69 |
+
chatbot 为WebUI中显示的对话列表,修改它,然后yeild出去,可以直接修改对话界面内容
|
| 70 |
+
additional_fn代表点击的哪个按钮,按钮见functional.py
|
| 71 |
+
"""
|
| 72 |
if additional_fn is not None:
|
| 73 |
import functional
|
| 74 |
importlib.reload(functional)
|
|
|
|
| 136 |
return
|
| 137 |
|
| 138 |
def generate_payload(inputs, top_p, temperature, history, system_prompt, stream):
|
| 139 |
+
"""
|
| 140 |
+
整合所有信息,选择LLM模型,生成http请求,为发送请求做准备
|
| 141 |
+
"""
|
| 142 |
headers = {
|
| 143 |
"Content-Type": "application/json",
|
| 144 |
"Authorization": f"Bearer {API_KEY}"
|
toolbox.py
CHANGED
|
@@ -10,7 +10,10 @@ def predict_no_ui_but_counting_down(i_say, i_say_show_user, chatbot, top_p, temp
|
|
| 10 |
try: from config_private import TIMEOUT_SECONDS, MAX_RETRY
|
| 11 |
except: from config import TIMEOUT_SECONDS, MAX_RETRY
|
| 12 |
from predict import predict_no_ui
|
|
|
|
|
|
|
| 13 |
mutable = [None, '']
|
|
|
|
| 14 |
def mt(i_say, history):
|
| 15 |
while True:
|
| 16 |
try:
|
|
@@ -25,14 +28,16 @@ def predict_no_ui_but_counting_down(i_say, i_say_show_user, chatbot, top_p, temp
|
|
| 25 |
mutable[1] = 'Warning! Input file is too long, cut into half. '
|
| 26 |
except TimeoutError as e:
|
| 27 |
mutable[0] = '[Local Message] Failed with timeout'
|
| 28 |
-
|
| 29 |
thread_name = threading.Thread(target=mt, args=(i_say, history)); thread_name.start()
|
|
|
|
| 30 |
cnt = 0
|
| 31 |
while thread_name.is_alive():
|
| 32 |
cnt += 1
|
| 33 |
chatbot[-1] = (i_say_show_user, f"[Local Message] {mutable[1]}waiting gpt response {cnt}/{TIMEOUT_SECONDS*2*(MAX_RETRY+1)}"+''.join(['.']*(cnt%4)))
|
| 34 |
yield chatbot, history, '正常'
|
| 35 |
time.sleep(1)
|
|
|
|
| 36 |
gpt_say = mutable[0]
|
| 37 |
return gpt_say
|
| 38 |
|
|
|
|
| 10 |
try: from config_private import TIMEOUT_SECONDS, MAX_RETRY
|
| 11 |
except: from config import TIMEOUT_SECONDS, MAX_RETRY
|
| 12 |
from predict import predict_no_ui
|
| 13 |
+
# 多线程的时候,需要一个mutable结构在不同线程之间传递信息
|
| 14 |
+
# list就是最简单的mutable结构,我们第一个位置放gpt输出,第二个位置传递报错信息
|
| 15 |
mutable = [None, '']
|
| 16 |
+
# multi-threading worker
|
| 17 |
def mt(i_say, history):
|
| 18 |
while True:
|
| 19 |
try:
|
|
|
|
| 28 |
mutable[1] = 'Warning! Input file is too long, cut into half. '
|
| 29 |
except TimeoutError as e:
|
| 30 |
mutable[0] = '[Local Message] Failed with timeout'
|
| 31 |
+
# 创建新线程发出http请求
|
| 32 |
thread_name = threading.Thread(target=mt, args=(i_say, history)); thread_name.start()
|
| 33 |
+
# 原来的线程则负责持续更新UI,实现一个超时倒计时,并等待新线程的任务完成
|
| 34 |
cnt = 0
|
| 35 |
while thread_name.is_alive():
|
| 36 |
cnt += 1
|
| 37 |
chatbot[-1] = (i_say_show_user, f"[Local Message] {mutable[1]}waiting gpt response {cnt}/{TIMEOUT_SECONDS*2*(MAX_RETRY+1)}"+''.join(['.']*(cnt%4)))
|
| 38 |
yield chatbot, history, '正常'
|
| 39 |
time.sleep(1)
|
| 40 |
+
# 把gpt的输出从mutable中取出来
|
| 41 |
gpt_say = mutable[0]
|
| 42 |
return gpt_say
|
| 43 |
|