Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -3,11 +3,8 @@ import multiprocessing
|
|
3 |
import concurrent.futures
|
4 |
from langchain.document_loaders import TextLoader, DirectoryLoader
|
5 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
6 |
-
from
|
7 |
-
from
|
8 |
-
import faiss
|
9 |
-
import torch
|
10 |
-
import numpy as np
|
11 |
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer, BitsAndBytesConfig
|
12 |
from datetime import datetime
|
13 |
import json
|
@@ -24,7 +21,7 @@ logger = logging.getLogger(__name__)
|
|
24 |
class DocumentRetrievalAndGeneration:
|
25 |
def __init__(self, embedding_model_name, lm_model_id, data_folder):
|
26 |
self.all_splits = self.load_documents(data_folder)
|
27 |
-
self.embeddings =
|
28 |
self.vectordb = self.create_faiss_index()
|
29 |
self.tokenizer, self.model = self.initialize_llm(lm_model_id)
|
30 |
self.retriever_tool = self.create_retriever_tool()
|
@@ -40,24 +37,7 @@ class DocumentRetrievalAndGeneration:
|
|
40 |
return all_splits
|
41 |
|
42 |
def create_faiss_index(self):
|
43 |
-
|
44 |
-
embeddings = self.embeddings.encode(all_texts)
|
45 |
-
|
46 |
-
# Create FAISS index
|
47 |
-
vector_dimension = embeddings.shape[1]
|
48 |
-
index = faiss.IndexFlatL2(vector_dimension)
|
49 |
-
index.add(embeddings)
|
50 |
-
|
51 |
-
# Create docstore
|
52 |
-
docstore = {i: doc for i, doc in enumerate(self.all_splits)}
|
53 |
-
|
54 |
-
# Create and return FAISS object
|
55 |
-
return FAISS(
|
56 |
-
embedding_function=self.embeddings.encode,
|
57 |
-
index=index,
|
58 |
-
docstore=docstore,
|
59 |
-
index_to_docstore_id={i: i for i in range(len(self.all_splits))}
|
60 |
-
)
|
61 |
|
62 |
def initialize_llm(self, model_id):
|
63 |
quantization_config = BitsAndBytesConfig(
|
@@ -145,6 +125,12 @@ Question:
|
|
145 |
response = self.query_and_generate_response(query)
|
146 |
return response
|
147 |
|
|
|
|
|
|
|
|
|
|
|
|
|
148 |
if __name__ == "__main__":
|
149 |
embedding_model_name = 'thenlper/gte-small'
|
150 |
lm_model_id = "meta-llama/Meta-Llama-3.1-8B-Instruct"
|
@@ -152,6 +138,9 @@ if __name__ == "__main__":
|
|
152 |
|
153 |
doc_retrieval_gen = DocumentRetrievalAndGeneration(embedding_model_name, lm_model_id, data_folder)
|
154 |
|
|
|
|
|
|
|
155 |
def launch_interface():
|
156 |
css_code = """
|
157 |
.gradio-container {
|
|
|
3 |
import concurrent.futures
|
4 |
from langchain.document_loaders import TextLoader, DirectoryLoader
|
5 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
6 |
+
from langchain_community.vectorstores import FAISS
|
7 |
+
from langchain_community.embeddings import HuggingFaceEmbeddings
|
|
|
|
|
|
|
8 |
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer, BitsAndBytesConfig
|
9 |
from datetime import datetime
|
10 |
import json
|
|
|
21 |
class DocumentRetrievalAndGeneration:
|
22 |
def __init__(self, embedding_model_name, lm_model_id, data_folder):
|
23 |
self.all_splits = self.load_documents(data_folder)
|
24 |
+
self.embeddings = HuggingFaceEmbeddings(model_name=embedding_model_name)
|
25 |
self.vectordb = self.create_faiss_index()
|
26 |
self.tokenizer, self.model = self.initialize_llm(lm_model_id)
|
27 |
self.retriever_tool = self.create_retriever_tool()
|
|
|
37 |
return all_splits
|
38 |
|
39 |
def create_faiss_index(self):
|
40 |
+
return FAISS.from_documents(self.all_splits, self.embeddings)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
|
42 |
def initialize_llm(self, model_id):
|
43 |
quantization_config = BitsAndBytesConfig(
|
|
|
125 |
response = self.query_and_generate_response(query)
|
126 |
return response
|
127 |
|
128 |
+
def save_index(self, path):
|
129 |
+
self.vectordb.save_local(path)
|
130 |
+
|
131 |
+
def load_index(self, path):
|
132 |
+
self.vectordb = FAISS.load_local(path, self.embeddings)
|
133 |
+
|
134 |
if __name__ == "__main__":
|
135 |
embedding_model_name = 'thenlper/gte-small'
|
136 |
lm_model_id = "meta-llama/Meta-Llama-3.1-8B-Instruct"
|
|
|
138 |
|
139 |
doc_retrieval_gen = DocumentRetrievalAndGeneration(embedding_model_name, lm_model_id, data_folder)
|
140 |
|
141 |
+
# Save the index for future use
|
142 |
+
doc_retrieval_gen.save_index("faiss_index")
|
143 |
+
|
144 |
def launch_interface():
|
145 |
css_code = """
|
146 |
.gradio-container {
|