FAISS vector db for HF spaces compatibility

fast_app.py

@@ -7,20 +7,15 @@ from fastapi.templating import Jinja2Templates
 from fastapi.staticfiles import StaticFiles
 from fastapi.encoders import jsonable_encoder
 
-from langchain.vectorstores import Chroma
+from langchain_community.vectorstores import FAISS
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 
 from langchain.chains import RetrievalQA
-from langchain.document_loaders import (
-    TextLoader,
-    PyPDFLoader,
-    DirectoryLoader,
-    UnstructuredFileLoader,
-)
-from langchain.document_loaders.csv_loader import CSVLoader
+
 from langchain.llms import OpenAI
 from langchain import PromptTemplate
 from langchain.embeddings import OpenAIEmbeddings, HuggingFaceEmbeddings
+import chainlit as cl
 
 from ingest import Ingest
 
@@ -31,7 +26,7 @@ from ingest import Ingest
 # if huggingface_token is None:
 #    raise ValueError("Hugging Face token is not set in environment variables.")
 
-openai_api_key = os.getenv("OPENAI_API_KEY")
+openai_api_key = "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"
 if openai_api_key is None:
     raise ValueError("OAI token is not set in environment variables.")
 
@@ -39,8 +34,8 @@ if openai_api_key is None:
 app = FastAPI()
 templates = Jinja2Templates(directory="templates")
 app.mount("/static", StaticFiles(directory="static"), name="static")
-english_embedding_model="text-embedding-3-large"
-czech_embedding_model="Seznam/simcse-dist-mpnet-paracrawl-cs-en"
+english_embedding_model = "text-embedding-3-large"
+czech_embedding_model = "Seznam/simcse-dist-mpnet-paracrawl-cs-en"
 
 czech_store = "stores/czech_512"
 english_store = "stores/english_512"
@@ -55,6 +50,7 @@ ingestor = Ingest(
     english_embedding_model=english_embedding_model,
 )
 
+
 def prompt_en():
     prompt_template_en = """You are electrical engineer and you answer users ###Question.
 
@@ -75,6 +71,7 @@ def prompt_en():
     print("\n Prompt ready... \n\n")
     return prompt_en
 
+
 def prompt_cz():
     prompt_template_cz = """Jste elektroinženýr a odpovídáte uživatelům na ###Otázku.
 
@@ -144,7 +141,7 @@ async def get_response(query: str = Form(...), language: str = Form(...)):
             model=embedding_model,
         )
 
-    vectordb = Chroma(persist_directory=persist_directory, embedding_function=embedding)
+    vectordb = FAISS.load_local(persist_directory, embedding)
     retriever = vectordb.as_retriever(search_kwargs={"k": 3})
 
     chain_type_kwargs = {"prompt": prompt}

ingest.py

@@ -1,4 +1,4 @@
-from langchain.vectorstores import Chroma
+from langchain_community.vectorstores import FAISS
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 
 from langchain.document_loaders import PyPDFLoader, DirectoryLoader
@@ -53,12 +53,11 @@ class Ingest:
         )
         texts = text_splitter.split_documents(documents)
 
-        vectordb = Chroma.from_documents(
+        vectordb = FAISS.from_documents(
             documents=texts,
             embedding=embedding,
-            persist_directory=self.english_store,
-            collection_metadata={"hnsw:space": "cosine"},
         )
+        vectordb.save_local(self.english_store)
 
         print("\n English vector Store Created.......\n\n")
 
@@ -84,12 +83,11 @@ class Ingest:
         )
 
         texts = text_splitter.split_documents(documents)
-        vectordb = Chroma.from_documents(
+        vectordb = FAISS.from_documents(
             documents=texts,
             embedding=embedding,
-            persist_directory=self.czech_store,
-            collection_metadata={"hnsw:space": "cosine"},
         )
+        vectordb.save_local(self.czech_store)
 
         print("\n Czech vector Store Created.......\n\n")
 

requirements.txt

@@ -1,5 +1,9 @@
-langchain
+
+langchain-community==0.0.19
+langchain==0.1.6
 fastapi
+faiss-cpu
+pypdf
 uvicorn
 python-multipart
 ctransformers
@@ -9,8 +13,8 @@ sentence_transformers
 chromadb
 pytesseract
 fitz
-libpff-python
+#libpff-python
 openai
 tiktoken
 frontend
-pysqlite3-binary
+chainlit

stores/czech_512/9b9472a9-9f91-4b34-880b-b7752517675a/data_level0.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5f8157971983f837eca48b97187f0e8a435eb21270cd49d831db21678670bc4a
-size 1164000

stores/czech_512/9b9472a9-9f91-4b34-880b-b7752517675a/header.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:9a3499aedbeb5c8ea26813ed567be6748293334099aa733c4d8cf0c4ec0ee6e3
-size 100

stores/czech_512/9b9472a9-9f91-4b34-880b-b7752517675a/length.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:612e017796cdd9eef6ba562cbe8c02e16b8c07f3fbac9f1254934f02e2261084
-size 4000

stores/czech_512/chroma.sqlite3

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2187862ccdfdb78565366853a939dc50038908171936c8584d69a09b55aa4e7c
-size 1929216

stores/english_512/3af5c10a-ea06-4cbe-beaf-8497680ad526/data_level0.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:6f812eacc9c05db367748cf1e0576bdcd28e0b3eaf09d5f3095a1b0e03f71cc8
-size 12428000

stores/english_512/3af5c10a-ea06-4cbe-beaf-8497680ad526/header.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:9882e5d786d4ca5fba4a783054685cf6e05b1637aaf586e43ec0e933e30e961d
-size 100

stores/english_512/3af5c10a-ea06-4cbe-beaf-8497680ad526/index_metadata.pickle

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d49c7e9538b2cfc154773a96a1fcdbf4a4247c3b510bb68d2aa6f2b24e902fca
-size 55974

stores/english_512/3af5c10a-ea06-4cbe-beaf-8497680ad526/length.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:bd6e73e535a8843ce30d35a4ba88436bcb5687583474e276a3b1f8689c1477bd
-size 4000

stores/english_512/3af5c10a-ea06-4cbe-beaf-8497680ad526/link_lists.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:fe35f087195e70122f597edc9b62da9d3ce370b40307b5556ebbe4e185fb46d4
-size 8624

stores/english_512/chroma.sqlite3

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:369ede691a1330113d353e1a425a7cd24ad9d76ee61ee542adab1f12a6887146
-size 26963968