Spaces:
Sleeping
Sleeping
Asaad Almutareb
commited on
Commit
•
45f1f60
1
Parent(s):
2675407
new repo Initial
Browse files- .devcontainer/Dockerfile +15 -0
- .devcontainer/devcontainer.json +37 -0
- .gitattributes +2 -0
- .github/workflows/check_file_size.yaml +16 -0
- .github/workflows/sync_2_hf.yml +20 -0
- .gitignore +8 -0
- README.md +1 -1
- app.py +137 -0
- example.env +1 -0
- localembedding.py +37 -0
- requirements.txt +8 -0
.devcontainer/Dockerfile
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
ARG VARIANT="3.10-bookworm"
|
2 |
+
#FROM mcr.microsoft.com/devcontainers/python:1-${VARIANT}
|
3 |
+
FROM langchain/langchain
|
4 |
+
|
5 |
+
# [Optional] Uncomment if you want to install an additional version of node using nvm
|
6 |
+
# ARG EXTRA_NODE_VERSION=10
|
7 |
+
# RUN su node -c "source /usr/local/share/nvm/nvm.sh && nvm install ${EXTRA_NODE_VERSION}"
|
8 |
+
|
9 |
+
# [Optional] Uncomment if you want to install more global node modules
|
10 |
+
# RUN su node -c "npm install -g <your-package-list-here>"
|
11 |
+
|
12 |
+
#COPY library-scripts/github-debian.sh /tmp/library-scripts/
|
13 |
+
RUN apt-get update && apt-get update
|
14 |
+
RUN pip install --upgrade pip
|
15 |
+
#RUN pip install -r requirements.txt
|
.devcontainer/devcontainer.json
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
// For format details, see https://aka.ms/devcontainer.json. For config options, see the
|
2 |
+
// README at: https://github.com/devcontainers/templates/tree/main/src/python
|
3 |
+
{
|
4 |
+
"name": "Langchain latest",
|
5 |
+
// Or use a Dockerfile or Docker Compose file. More info: https://containers.dev/guide/dockerfile
|
6 |
+
//"image": "mcr.microsoft.com/devcontainers/python:1-3.10-bookworm",
|
7 |
+
|
8 |
+
// build config for the docker image instead:
|
9 |
+
"build": { "dockerfile": "Dockerfile" },
|
10 |
+
|
11 |
+
// Features to add to the dev container. More info: https://containers.dev/features.
|
12 |
+
// "features": {},
|
13 |
+
|
14 |
+
// Use 'forwardPorts' to make a list of ports inside the container available locally.
|
15 |
+
// "forwardPorts": [],
|
16 |
+
|
17 |
+
// Use 'postCreateCommand' to run commands after the container is created.
|
18 |
+
// "postCreateCommand": "pip3 install --user -r requirements.txt",
|
19 |
+
|
20 |
+
// Configure tool-specific properties.
|
21 |
+
"customizations": {
|
22 |
+
// Configure properties specific to VS Code.
|
23 |
+
"vscode": {
|
24 |
+
//Add the IDs of extensions you want installed when the container is created.
|
25 |
+
"extensions": ["ms-azuretools.vscode-docker"]
|
26 |
+
}
|
27 |
+
} //,
|
28 |
+
|
29 |
+
// Uncomment to connect as root instead. More info: https://aka.ms/dev-containers-non-root.
|
30 |
+
// "remoteUser": "root"
|
31 |
+
|
32 |
+
// mount the SSH directory on win
|
33 |
+
/*"mounts": [
|
34 |
+
"type=bind,source=${localWorkspaceFolder},target=/work",
|
35 |
+
"type=bind,source=/home/${localEnv:USER}/.ssh,target=/home/vscode/.ssh,readonly"
|
36 |
+
]*/
|
37 |
+
}
|
.gitattributes
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
*.sqlite3 filter=lfs diff=lfs merge=lfs -text
|
2 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
.github/workflows/check_file_size.yaml
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
name: Check file size
|
2 |
+
on: # or directly `on: [push]` to run the action on every push on any branch
|
3 |
+
pull_request:
|
4 |
+
branches: [main]
|
5 |
+
|
6 |
+
# to run this workflow manually from the Actions tab
|
7 |
+
workflow_dispatch:
|
8 |
+
|
9 |
+
jobs:
|
10 |
+
sync-to-hub:
|
11 |
+
runs-on: ubuntu-latest
|
12 |
+
steps:
|
13 |
+
- name: Check large files
|
14 |
+
uses: ActionsDesk/lfs-warning@v2.0
|
15 |
+
with:
|
16 |
+
filesizelimit: 10485760 # this is 10MB so we can sync to HF Spaces
|
.github/workflows/sync_2_hf.yml
ADDED
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
name: Sync to Hugging Face hub
|
2 |
+
on:
|
3 |
+
push:
|
4 |
+
branches: [main]
|
5 |
+
|
6 |
+
# to run this workflow manually from the Actions tab
|
7 |
+
workflow_dispatch:
|
8 |
+
|
9 |
+
jobs:
|
10 |
+
sync-to-hub:
|
11 |
+
runs-on: ubuntu-latest
|
12 |
+
steps:
|
13 |
+
- uses: actions/checkout@v3
|
14 |
+
with:
|
15 |
+
fetch-depth: 0
|
16 |
+
lfs: true
|
17 |
+
- name: Push to hub
|
18 |
+
env:
|
19 |
+
HF_TOKEN: ${{ secrets.HF_TOKEN }}
|
20 |
+
run: git push https://sabazo:$HF_TOKEN@huggingface.co/spaces/sabazo/docs-qachat-demo main
|
.gitignore
CHANGED
@@ -158,3 +158,11 @@ cython_debug/
|
|
158 |
# and can be added to the global gitignore or merged into this file. For a more nuclear
|
159 |
# option (not recommended) you can uncomment the following to ignore the entire idea folder.
|
160 |
#.idea/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
158 |
# and can be added to the global gitignore or merged into this file. For a more nuclear
|
159 |
# option (not recommended) you can uncomment the following to ignore the entire idea folder.
|
160 |
#.idea/
|
161 |
+
|
162 |
+
# embeddings db
|
163 |
+
*.sqlite3*
|
164 |
+
*.bin
|
165 |
+
*.pickle
|
166 |
+
chroma_db/*
|
167 |
+
bin
|
168 |
+
obj
|
README.md
CHANGED
@@ -1 +1 @@
|
|
1 |
-
#
|
|
|
1 |
+
# docu-qachat-demo
|
app.py
ADDED
@@ -0,0 +1,137 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# gradio
|
2 |
+
import gradio as gr
|
3 |
+
import random
|
4 |
+
import time
|
5 |
+
#boto3 for S3 access
|
6 |
+
import boto3
|
7 |
+
from botocore import UNSIGNED
|
8 |
+
from botocore.client import Config
|
9 |
+
# access .env file
|
10 |
+
from dotenv import load_dotenv
|
11 |
+
#from bs4 import BeautifulSoup
|
12 |
+
# HF libraries
|
13 |
+
from langchain.llms import HuggingFaceHub
|
14 |
+
from langchain.embeddings import HuggingFaceHubEmbeddings
|
15 |
+
# vectorestore
|
16 |
+
from langchain.vectorstores import Chroma
|
17 |
+
from langchain.vectorstores import FAISS
|
18 |
+
# retrieval chain
|
19 |
+
from langchain.chains import RetrievalQA
|
20 |
+
# prompt template
|
21 |
+
from langchain.prompts import PromptTemplate
|
22 |
+
from langchain.memory import ConversationBufferMemory
|
23 |
+
# logging
|
24 |
+
#import logging
|
25 |
+
import zipfile
|
26 |
+
|
27 |
+
# load HF Token
|
28 |
+
config = load_dotenv(".env")
|
29 |
+
|
30 |
+
|
31 |
+
model_id = HuggingFaceHub(repo_id="HuggingFaceH4/zephyr-7b-beta", model_kwargs={
|
32 |
+
"temperature":0.1,
|
33 |
+
"max_new_tokens":1024,
|
34 |
+
"repetition_penalty":1.2,
|
35 |
+
"streaming": True,
|
36 |
+
"return_full_text":True
|
37 |
+
})
|
38 |
+
|
39 |
+
model_name = "sentence-transformers/multi-qa-mpnet-base-dot-v1"
|
40 |
+
embeddings = HuggingFaceHubEmbeddings(repo_id=model_name)
|
41 |
+
|
42 |
+
|
43 |
+
s3 = boto3.client('s3', config=Config(signature_version=UNSIGNED))
|
44 |
+
|
45 |
+
## Chroma DB
|
46 |
+
s3.download_file('rad-rag-demos', 'vectorstores/chroma.sqlite3', './chroma_db/chroma.sqlite3')
|
47 |
+
db = Chroma(persist_directory="./chroma_db", embedding_function=embeddings)
|
48 |
+
db.get()
|
49 |
+
|
50 |
+
## FAISS DB
|
51 |
+
# s3.download_file('rad-rag-demos', 'vectorstores/faiss_db_ray.zip', './chroma_db/faiss_db_ray.zip')
|
52 |
+
# with zipfile.ZipFile('./chroma_db/faiss_db_ray.zip', 'r') as zip_ref:
|
53 |
+
# zip_ref.extractall('./chroma_db/')
|
54 |
+
|
55 |
+
# FAISS_INDEX_PATH='./chroma_db/faiss_db_ray'
|
56 |
+
# db = FAISS.load_local(FAISS_INDEX_PATH, embeddings)
|
57 |
+
|
58 |
+
retriever = db.as_retriever(search_type = "mmr")#, search_kwargs={'k': 5, 'fetch_k': 25})
|
59 |
+
global qa
|
60 |
+
template = """
|
61 |
+
You are the friendly documentation buddy Arti, who helps the Human in using RAY, the open-source unified framework for scaling AI and Python applications.\
|
62 |
+
Use the following context (delimited by <ctx></ctx>) and the chat history (delimited by <hs></hs>) to answer the question :
|
63 |
+
------
|
64 |
+
<ctx>
|
65 |
+
{context}
|
66 |
+
</ctx>
|
67 |
+
------
|
68 |
+
<hs>
|
69 |
+
{history}
|
70 |
+
</hs>
|
71 |
+
------
|
72 |
+
{question}
|
73 |
+
Answer:
|
74 |
+
"""
|
75 |
+
prompt = PromptTemplate(
|
76 |
+
input_variables=["history", "context", "question"],
|
77 |
+
template=template,
|
78 |
+
)
|
79 |
+
memory = ConversationBufferMemory(memory_key="history", input_key="question")
|
80 |
+
qa = RetrievalQA.from_chain_type(llm=model_id, chain_type="stuff", retriever=retriever, verbose=True, return_source_documents=True, chain_type_kwargs={
|
81 |
+
"verbose": True,
|
82 |
+
"memory": memory,
|
83 |
+
"prompt": prompt
|
84 |
+
}
|
85 |
+
)
|
86 |
+
|
87 |
+
|
88 |
+
def add_text(history, text):
|
89 |
+
history = history + [(text, None)]
|
90 |
+
return history, ""
|
91 |
+
|
92 |
+
def bot(history):
|
93 |
+
response = infer(history[-1][0], history)
|
94 |
+
print(*memory)
|
95 |
+
sources = [doc.metadata.get("source") for doc in response['source_documents']]
|
96 |
+
src_list = '\n'.join(sources)
|
97 |
+
print_this = response['result']+"\n\n\n Sources: \n\n\n"+src_list
|
98 |
+
|
99 |
+
#history[-1][1] = ""
|
100 |
+
#for character in response['result']: #print_this:
|
101 |
+
# history[-1][1] += character
|
102 |
+
# time.sleep(0.05)
|
103 |
+
# yield history
|
104 |
+
history[-1][1] = print_this #response['result']
|
105 |
+
return history
|
106 |
+
|
107 |
+
def infer(question, history):
|
108 |
+
query = question
|
109 |
+
result = qa({"query": query, "history": history, "question": question})
|
110 |
+
return result
|
111 |
+
|
112 |
+
css="""
|
113 |
+
#col-container {max-width: 700px; margin-left: auto; margin-right: auto;}
|
114 |
+
"""
|
115 |
+
|
116 |
+
title = """
|
117 |
+
<div style="text-align: center;max-width: 700px;">
|
118 |
+
<h1>Chat with your Documentation</h1>
|
119 |
+
<p style="text-align: center;">Chat with Documentation, <br />
|
120 |
+
when everything is ready, you can start asking questions about the docu ;)</p>
|
121 |
+
</div>
|
122 |
+
"""
|
123 |
+
|
124 |
+
with gr.Blocks(css=css) as demo:
|
125 |
+
with gr.Column(elem_id="col-container"):
|
126 |
+
gr.HTML(title)
|
127 |
+
chatbot = gr.Chatbot([], elem_id="chatbot")
|
128 |
+
clear = gr.Button("Clear")
|
129 |
+
with gr.Row():
|
130 |
+
question = gr.Textbox(label="Question", placeholder="Type your question and hit Enter ")
|
131 |
+
question.submit(add_text, [chatbot, question], [chatbot, question], queue=False).then(
|
132 |
+
bot, chatbot, chatbot
|
133 |
+
)
|
134 |
+
clear.click(lambda: None, None, chatbot, queue=False)
|
135 |
+
|
136 |
+
demo.queue()
|
137 |
+
demo.launch()
|
example.env
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
OPENAI_API_KEY=
|
localembedding.py
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import chromadb
|
2 |
+
from chromadb.config import Settings
|
3 |
+
from langchain.vectorstores import Chroma
|
4 |
+
from langchain.vectorstores.utils import filter_complex_metadata
|
5 |
+
import time
|
6 |
+
from langchain.embeddings import HuggingFaceEmbeddings
|
7 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
8 |
+
|
9 |
+
text_splitter = RecursiveCharacterTextSplitter(
|
10 |
+
# Set a really small chunk size, just to show.
|
11 |
+
chunk_size = 1000,
|
12 |
+
chunk_overlap = 100,
|
13 |
+
length_function = len,
|
14 |
+
)
|
15 |
+
|
16 |
+
# Stage one: read all the docs, split them into chunks.
|
17 |
+
st = time.time()
|
18 |
+
print('Loading documents ...')
|
19 |
+
docs = loader.load()
|
20 |
+
chunks = text_splitter.create_documents([doc.page_content for doc in docs], metadatas=[doc.metadata for doc in docs])
|
21 |
+
et = time.time() - st
|
22 |
+
print(f'Time taken: {et} seconds.')
|
23 |
+
|
24 |
+
#Stage two: embed the docs.
|
25 |
+
# use all-mpnet-base-v2 sentence transformer to convert pieces of text in vectors to store them in the vector store
|
26 |
+
model_name = "sentence-transformers/all-mpnet-base-v2"
|
27 |
+
model_kwargs = {"device": "cuda"}
|
28 |
+
|
29 |
+
embeddings = HuggingFaceEmbeddings(
|
30 |
+
model_name=model_name,
|
31 |
+
model_kwargs=model_kwargs
|
32 |
+
)
|
33 |
+
print(f'Loading chunks into vector store ...')
|
34 |
+
st = time.time()
|
35 |
+
db = Chroma.from_documents(filter_complex_metadata(chunks), embeddings, persist_directory="/content/chroma_db")
|
36 |
+
et = time.time() - st
|
37 |
+
print(f'Time taken: {et} seconds.')
|
requirements.txt
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
huggingface_hub
|
2 |
+
chromadb
|
3 |
+
faiss-cpu
|
4 |
+
langchain
|
5 |
+
python-dotenv
|
6 |
+
bs4
|
7 |
+
gradio
|
8 |
+
boto3
|