github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-03 18:19:58 +00:00
Code Issues Packages Projects Releases Wiki Activity

[devops] remove post commit ci (#5566)

Browse Source 
* [devops] remove post commit ci

* [misc] run pre-commit on all files

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
This commit is contained in:
Hongxin Liu
2024-04-08 15:09:40 +08:00
committed by GitHub
parent 341263df48
commit 641b1ee71a
82 changed files with 849 additions and 962 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
applications/ColossalQA/examples/webui_demo/RAG_ChatBot.py
View File

@@ -5,13 +5,7 @@ from colossalqa.chain.retrieval_qa.base import RetrievalQA
from colossalqa.data_loader.document_loader import DocumentLoader
from colossalqa.memory import ConversationBufferWithSummary
from colossalqa.mylogging import get_logger
from colossalqa.prompt.prompt import (
PROMPT_DISAMBIGUATE_ZH,
PROMPT_RETRIEVAL_QA_ZH,
SUMMARY_PROMPT_ZH,
ZH_RETRIEVAL_QA_REJECTION_ANSWER,
ZH_RETRIEVAL_QA_TRIGGER_KEYWORDS,
)
from colossalqa.prompt.prompt import ZH_RETRIEVAL_QA_REJECTION_ANSWER, ZH_RETRIEVAL_QA_TRIGGER_KEYWORDS
from colossalqa.retriever import CustomRetriever
from langchain import LLMChain
from langchain.embeddings import HuggingFaceEmbeddings
@@ -116,13 +110,13 @@ class RAG_ChatBot:
def split_docs(self, documents):
doc_splits = self.text_splitter.split_documents(documents)
return doc_splits
def clear_docs(self, **kwargs):
self.documents = []
self.docs_names = []
self.info_retriever.clear_documents()
self.memory.initiate_document_retrieval_chain(self.llm, kwargs["gen_qa_prompt"], self.info_retriever)
def reset_config(self, rag_config):
self.rag_config = rag_config
self.set_embed_model(**self.rag_config["embed"])

2
applications/ColossalQA/examples/webui_demo/README.md
View File

@@ -115,4 +115,4 @@ python webui.py --http_host "your-backend-api-host" --http_port "your-backend-ap
After launching the script, you can upload files and engage with the chatbot through your web browser.
![ColossalQA Demo](https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/colossalqa/new_ui.png)
![ColossalQA Demo](https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/colossalqa/new_ui.png)

52
applications/ColossalQA/examples/webui_demo/config.py
View File

@@ -1,58 +1,30 @@
from colossalqa.prompt.prompt import (
PROMPT_DISAMBIGUATE_ZH,
PROMPT_RETRIEVAL_QA_ZH,
SUMMARY_PROMPT_ZH,
ZH_RETRIEVAL_QA_REJECTION_ANSWER,
ZH_RETRIEVAL_QA_TRIGGER_KEYWORDS,
)
from colossalqa.prompt.prompt import PROMPT_DISAMBIGUATE_ZH, PROMPT_RETRIEVAL_QA_ZH, SUMMARY_PROMPT_ZH
from colossalqa.text_splitter import ChineseTextSplitter
ALL_CONFIG = {
"embed": {
"embed_name": "m3e", # embedding model name
"embed_model_name_or_path": "moka-ai/m3e-base", # path to embedding model, could be a local path or a huggingface path
"embed_model_device": {
"device": "cpu"
}
"embed_model_device": {"device": "cpu"},
},
"model": {
"mode": "api", # "local" for loading models, "api" for using model api
"model_name": "chatgpt_api", # local model name, "chatgpt_api" or "pangu_api"
"model_path": "", # path to the model, could be a local path or a huggingface path. don't need if using an api
"device": {
"device": "cuda"
}
},
"splitter": {
"name": ChineseTextSplitter
},
"retrieval": {
"retri_top_k": 3,
"retri_kb_file_path": "./", # path to store database files
"verbose": True
"model_path": "", # path to the model, could be a local path or a huggingface path. don't need if using an api
"device": {"device": "cuda"},
},
"splitter": {"name": ChineseTextSplitter},
"retrieval": {"retri_top_k": 3, "retri_kb_file_path": "./", "verbose": True}, # path to store database files
"chain": {
"mem_summary_prompt": SUMMARY_PROMPT_ZH, # summary prompt template
"mem_human_prefix": "用户",
"mem_ai_prefix": "Assistant",
"mem_max_tokens": 2000,
"mem_llm_kwargs": {
"max_new_tokens": 50,
"temperature": 1,
"do_sample": True
},
"mem_llm_kwargs": {"max_new_tokens": 50, "temperature": 1, "do_sample": True},
"disambig_prompt": PROMPT_DISAMBIGUATE_ZH, # disambiguate prompt template
"disambig_llm_kwargs": {
"max_new_tokens": 30,
"temperature": 1,
"do_sample": True
},
"gen_llm_kwargs": {
"max_new_tokens": 100,
"temperature": 1,
"do_sample": True
},
"disambig_llm_kwargs": {"max_new_tokens": 30, "temperature": 1, "do_sample": True},
"gen_llm_kwargs": {"max_new_tokens": 100, "temperature": 1, "do_sample": True},
"gen_qa_prompt": PROMPT_RETRIEVAL_QA_ZH, # generation prompt template
"verbose": True
}
}
"verbose": True,
},
}

29
applications/ColossalQA/examples/webui_demo/server.py
View File

@@ -1,27 +1,18 @@
import argparse
import os
from typing import List, Union
from colossalqa.local.llm import ColossalAPI, ColossalLLM
from colossalqa.data_loader.document_loader import DocumentLoader
from colossalqa.mylogging import get_logger
from colossalqa.retrieval_conversation_zh import ChineseRetrievalConversation
from colossalqa.retriever import CustomRetriever
from enum import Enum
from fastapi import FastAPI, Request
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from pydantic import BaseModel, Field
import uvicorn
import config
import uvicorn
from colossalqa.local.llm import ColossalAPI, ColossalLLM
from colossalqa.mylogging import get_logger
from fastapi import FastAPI, Request
from pydantic import BaseModel
from RAG_ChatBot import RAG_ChatBot
from utils import DocAction
logger = get_logger()
def parseArgs():
parser = argparse.ArgumentParser()
parser.add_argument("--http_host", default="0.0.0.0")
@@ -36,6 +27,7 @@ class DocUpdateReq(BaseModel):
doc_files: Union[List[str], str, None] = None
action: DocAction = DocAction.ADD
class GenerationTaskReq(BaseModel):
user_input: str
@@ -45,7 +37,7 @@ def update_docs(data: DocUpdateReq, request: Request):
if data.action == "add":
if isinstance(data.doc_files, str):
data.doc_files = [data.doc_files]
chatbot.load_doc_from_files(files = data.doc_files)
chatbot.load_doc_from_files(files=data.doc_files)
all_docs = ""
for doc in chatbot.docs_names:
all_docs += f"\t{doc}\n\n"
@@ -79,17 +71,18 @@ if __name__ == "__main__":
elif all_config["model"]["mode"] == "api":
if model_name == "pangu_api":
from colossalqa.local.pangu_llm import Pangu
gen_config = {
"user": "User",
"max_tokens": all_config["chain"]["disambig_llm_kwargs"]["max_new_tokens"],
"temperature": all_config["chain"]["disambig_llm_kwargs"]["temperature"],
"n": 1 # the number of responses generated
"n": 1, # the number of responses generated
}
llm = Pangu(gen_config=gen_config)
llm.set_auth_config() # verify user's auth info here
elif model_name == "chatgpt_api":
from langchain.llms import OpenAI
llm = OpenAI()
else:
raise ValueError("Unsupported mode.")

31
applications/ColossalQA/examples/webui_demo/webui.py
View File

@@ -1,24 +1,26 @@
import argparse
import json
import os
import requests
import gradio as gr
import requests
from utils import DocAction
def parseArgs():
parser = argparse.ArgumentParser()
parser.add_argument("--http_host", default="0.0.0.0")
parser.add_argument("--http_port", type=int, default=13666)
return parser.parse_args()
def get_response(data, url):
headers = {"Content-type": "application/json"}
response = requests.post(url, json=data, headers=headers)
response = json.loads(response.content)
return response
def add_text(history, text):
history = history + [(text, None)]
return history, gr.update(value=None, interactive=True)
@@ -28,35 +30,28 @@ def add_file(history, files):
files_string = "\n".join([os.path.basename(file.name) for file in files])
doc_files = [file.name for file in files]
data = {
"doc_files": doc_files,
"action": DocAction.ADD
}
data = {"doc_files": doc_files, "action": DocAction.ADD}
response = get_response(data, update_url)["response"]
history = history + [(files_string, response)]
return history
def bot(history):
data = {
"user_input": history[-1][0].strip()
}
def bot(history):
data = {"user_input": history[-1][0].strip()}
response = get_response(data, gen_url)
if response["error"] != "":
raise gr.Error(response["error"])
history[-1][1] = response["response"]
yield history
def restart(chatbot, txt):
# Reset the conversation state and clear the chat history
data = {
"doc_files": "",
"action": DocAction.CLEAR
}
response = get_response(data, update_url)
data = {"doc_files": "", "action": DocAction.CLEAR}
get_response(data, update_url)
return gr.update(value=None), gr.update(value=None, interactive=True)
@@ -97,7 +92,7 @@ with gr.Blocks(css=CSS) as demo:
txt_msg = txt.submit(add_text, [chatbot, txt], [chatbot, txt], queue=False).then(bot, chatbot, chatbot)
# Clear the original textbox
txt_msg.then(lambda: gr.update(value=None, interactive=True), None, [txt], queue=False)
txt_msg.then(lambda: gr.update(value=None, interactive=True), None, [txt], queue=False)
# Click Upload Button: 1. upload files 2. send config to backend, initalize model 3. get response "conversation_ready" = True/False
file_msg = btn.upload(add_file, [chatbot, btn], [chatbot], queue=False)