Spaces:
Sleeping
Sleeping
Upload 2 files
Browse files- app.py +52 -23
- requirements.txt +8 -4
app.py
CHANGED
@@ -14,6 +14,11 @@ import chainlit as cl
|
|
14 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
15 |
# from langchain_experimental.text_splitter import SemanticChunker
|
16 |
# from langchain_openai.embeddings import OpenAIEmbeddings
|
|
|
|
|
|
|
|
|
|
|
17 |
|
18 |
system_template = """\
|
19 |
Use the following context to answer a users question. If you cannot find the answer in the context, say you don't know the answer."""
|
@@ -27,27 +32,27 @@ Question:
|
|
27 |
"""
|
28 |
user_role_prompt = UserRolePrompt(user_prompt_template)
|
29 |
|
30 |
-
class RetrievalAugmentedQAPipeline:
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
|
35 |
-
|
36 |
-
|
37 |
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
|
42 |
-
|
43 |
|
44 |
-
|
45 |
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
|
50 |
-
|
51 |
|
52 |
text_splitter = RecursiveCharacterTextSplitter()
|
53 |
|
@@ -90,6 +95,7 @@ async def on_chat_start():
|
|
90 |
max_files=10
|
91 |
).send()
|
92 |
|
|
|
93 |
for file in files:
|
94 |
|
95 |
msg = cl.Message(
|
@@ -99,26 +105,49 @@ async def on_chat_start():
|
|
99 |
|
100 |
# load the file
|
101 |
texts = process_text_file(file)
|
102 |
-
|
103 |
print(f"Processing {len(texts)} text chunks")
|
104 |
|
105 |
# Create a dict vector store
|
106 |
-
vector_db = VectorDatabase()
|
107 |
-
vector_db = await vector_db.abuild_from_list(texts)
|
108 |
|
109 |
-
chat_openai = ChatOpenAI()
|
110 |
|
111 |
# Create a chain
|
112 |
-
retrieval_augmented_qa_pipeline = RetrievalAugmentedQAPipeline(
|
113 |
-
|
114 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
)
|
|
|
116 |
|
117 |
# Let the user know that the system is ready
|
118 |
msg.content = f"Processing `{file.name}` done. You can now ask questions!"
|
119 |
await msg.update()
|
120 |
|
121 |
-
cl.user_session.set("chain",
|
122 |
|
123 |
|
124 |
@cl.on_message
|
|
|
14 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
15 |
# from langchain_experimental.text_splitter import SemanticChunker
|
16 |
# from langchain_openai.embeddings import OpenAIEmbeddings
|
17 |
+
from sentence_transformers import SentenceTransformer
|
18 |
+
from langchain_huggingface import HuggingFaceEmbeddings
|
19 |
+
from langchain_community.vectorstores import FAISS
|
20 |
+
from langchain_openai.embeddings import OpenAIEmbeddings
|
21 |
+
from langchain_core.documents import Document
|
22 |
|
23 |
system_template = """\
|
24 |
Use the following context to answer a users question. If you cannot find the answer in the context, say you don't know the answer."""
|
|
|
32 |
"""
|
33 |
user_role_prompt = UserRolePrompt(user_prompt_template)
|
34 |
|
35 |
+
# class RetrievalAugmentedQAPipeline:
|
36 |
+
# def __init__(self, llm: ChatOpenAI(), vector_db_retriever: VectorDatabase) -> None:
|
37 |
+
# self.llm = llm
|
38 |
+
# self.vector_db_retriever = vector_db_retriever
|
39 |
|
40 |
+
# async def arun_pipeline(self, user_query: str):
|
41 |
+
# context_list = self.vector_db_retriever.search_by_text(user_query, k=4)
|
42 |
|
43 |
+
# context_prompt = ""
|
44 |
+
# for context in context_list:
|
45 |
+
# context_prompt += context[0] + "\n"
|
46 |
|
47 |
+
# formatted_system_prompt = system_role_prompt.create_message()
|
48 |
|
49 |
+
# formatted_user_prompt = user_role_prompt.create_message(question=user_query, context=context_prompt)
|
50 |
|
51 |
+
# async def generate_response():
|
52 |
+
# async for chunk in self.llm.astream([formatted_system_prompt, formatted_user_prompt]):
|
53 |
+
# yield chunk
|
54 |
|
55 |
+
# return {"response": generate_response(), "context": context_list}
|
56 |
|
57 |
text_splitter = RecursiveCharacterTextSplitter()
|
58 |
|
|
|
95 |
max_files=10
|
96 |
).send()
|
97 |
|
98 |
+
processed_documents = []
|
99 |
for file in files:
|
100 |
|
101 |
msg = cl.Message(
|
|
|
105 |
|
106 |
# load the file
|
107 |
texts = process_text_file(file)
|
108 |
+
processed_documents.extend(texts)
|
109 |
print(f"Processing {len(texts)} text chunks")
|
110 |
|
111 |
# Create a dict vector store
|
112 |
+
# vector_db = VectorDatabase()
|
113 |
+
# vector_db = await vector_db.abuild_from_list(texts)
|
114 |
|
115 |
+
# chat_openai = ChatOpenAI()
|
116 |
|
117 |
# Create a chain
|
118 |
+
# retrieval_augmented_qa_pipeline = RetrievalAugmentedQAPipeline(
|
119 |
+
# vector_db_retriever=vector_db,
|
120 |
+
# llm=chat_openai
|
121 |
+
# )
|
122 |
+
|
123 |
+
|
124 |
+
|
125 |
+
finetune_embeddings = HuggingFaceEmbeddings(model_name="finetuned_arctic")
|
126 |
+
|
127 |
+
finetune_vectorstore = FAISS.from_documents(processed_documents, finetune_embeddings)
|
128 |
+
finetune_retriever = finetune_vectorstore.as_retriever(search_kwargs={"k": 6})
|
129 |
+
|
130 |
+
from operator import itemgetter
|
131 |
+
from langchain_core.output_parsers import StrOutputParser
|
132 |
+
from langchain_core.runnables import RunnablePassthrough, RunnableParallel
|
133 |
+
|
134 |
+
rag_llm = ChatOpenAI(
|
135 |
+
model="gpt-4o-mini",
|
136 |
+
temperature=0
|
137 |
+
)
|
138 |
+
|
139 |
+
finetune_rag_chain = (
|
140 |
+
{"context": itemgetter("question") | finetune_retriever, "question": itemgetter("question")}
|
141 |
+
| RunnablePassthrough.assign(context=itemgetter("context"))
|
142 |
+
| {"response": system_template | rag_llm | StrOutputParser(), "context": itemgetter("context")}
|
143 |
)
|
144 |
+
|
145 |
|
146 |
# Let the user know that the system is ready
|
147 |
msg.content = f"Processing `{file.name}` done. You can now ask questions!"
|
148 |
await msg.update()
|
149 |
|
150 |
+
cl.user_session.set("chain", finetune_rag_chain)
|
151 |
|
152 |
|
153 |
@cl.on_message
|
requirements.txt
CHANGED
@@ -1,7 +1,11 @@
|
|
1 |
numpy
|
2 |
chainlit==0.7.700
|
3 |
-
openai
|
4 |
-
langchain_community
|
5 |
-
langchain_experimental
|
6 |
-
langchain_openai
|
|
|
|
|
|
|
|
|
7 |
pypdf
|
|
|
1 |
numpy
|
2 |
chainlit==0.7.700
|
3 |
+
# openai
|
4 |
+
# langchain_community
|
5 |
+
# langchain_experimental
|
6 |
+
# langchain_openai
|
7 |
+
# langchain_huggingface
|
8 |
+
langchain-core==0.2.40
|
9 |
+
langchain-openai==0.1.25
|
10 |
+
langchain-huggingface==0.0.3
|
11 |
pypdf
|