Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -14,12 +14,11 @@ from langchain.llms import HuggingFaceHub, LlamaCpp
|
|
14 |
from huggingface_hub import snapshot_download, hf_hub_download
|
15 |
|
16 |
|
17 |
-
#
|
|
|
|
|
18 |
|
19 |
-
repo_name =
|
20 |
-
model_name = "model-q4_K.gguf"
|
21 |
-
|
22 |
-
#snapshot_download(repo_id=repo_name, local_dir=".", allow_patterns=model_name)
|
23 |
|
24 |
|
25 |
def get_pdf_text(pdf_docs):
|
@@ -34,8 +33,8 @@ def get_pdf_text(pdf_docs):
|
|
34 |
|
35 |
def get_text_chunks(text):
|
36 |
text_splitter = CharacterTextSplitter(separator="\n",
|
37 |
-
chunk_size=1000,
|
38 |
-
chunk_overlap=200,
|
39 |
length_function=len
|
40 |
)
|
41 |
chunks = text_splitter.split_text(text)
|
@@ -44,10 +43,7 @@ def get_text_chunks(text):
|
|
44 |
|
45 |
|
46 |
def get_vectorstore(text_chunks):
|
47 |
-
#embeddings = OpenAIEmbeddings()
|
48 |
-
#embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
|
49 |
embeddings = HuggingFaceEmbeddings(model_name="intfloat/multilingual-e5-large")
|
50 |
-
#embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-mpnet-base-v2")
|
51 |
vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
|
52 |
|
53 |
return vectorstore
|
@@ -65,8 +61,6 @@ def get_conversation_chain(vectorstore, model_name):
|
|
65 |
# echo=True
|
66 |
# )
|
67 |
|
68 |
-
# llm = ChatOpenAI()
|
69 |
-
|
70 |
llm = GigaChat(credentials=os.getenv("GIGACHAT_CREDENTIALS"),
|
71 |
verify_ssl_certs=False)
|
72 |
|
@@ -106,13 +100,6 @@ def handle_userinput(user_question):
|
|
106 |
st.write(bot_template.replace(
|
107 |
"{{MSG}}", str(text.page_content)), unsafe_allow_html=True)
|
108 |
|
109 |
-
|
110 |
-
|
111 |
-
#for text in enumerate(st.session_state.retrieved_text):
|
112 |
-
# st.write(text[1].page_content, '\n')
|
113 |
-
|
114 |
-
#print(response['source_documents'][0])
|
115 |
-
|
116 |
# main code
|
117 |
load_dotenv()
|
118 |
|
|
|
14 |
from huggingface_hub import snapshot_download, hf_hub_download
|
15 |
|
16 |
|
17 |
+
# If you want to use gguf model, uncomment 18-19 and 54-62 lines, comment-out 64-65. Otherwise provide GigaChat Credentials through HF secrets menu
|
18 |
+
#repo_name = "IlyaGusev/saiga_mistral_7b_gguf"
|
19 |
+
#model_name = "model-q4_K.gguf"
|
20 |
|
21 |
+
snapshot_download(repo_id=repo_name, local_dir=".", allow_patterns=model_name)
|
|
|
|
|
|
|
22 |
|
23 |
|
24 |
def get_pdf_text(pdf_docs):
|
|
|
33 |
|
34 |
def get_text_chunks(text):
|
35 |
text_splitter = CharacterTextSplitter(separator="\n",
|
36 |
+
chunk_size=1000,
|
37 |
+
chunk_overlap=200,
|
38 |
length_function=len
|
39 |
)
|
40 |
chunks = text_splitter.split_text(text)
|
|
|
43 |
|
44 |
|
45 |
def get_vectorstore(text_chunks):
|
|
|
|
|
46 |
embeddings = HuggingFaceEmbeddings(model_name="intfloat/multilingual-e5-large")
|
|
|
47 |
vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
|
48 |
|
49 |
return vectorstore
|
|
|
61 |
# echo=True
|
62 |
# )
|
63 |
|
|
|
|
|
64 |
llm = GigaChat(credentials=os.getenv("GIGACHAT_CREDENTIALS"),
|
65 |
verify_ssl_certs=False)
|
66 |
|
|
|
100 |
st.write(bot_template.replace(
|
101 |
"{{MSG}}", str(text.page_content)), unsafe_allow_html=True)
|
102 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
# main code
|
104 |
load_dotenv()
|
105 |
|