Spaces:
Running
Running
Commit
·
3dff4cb
1
Parent(s):
be312e0
Updating Chat bot
Browse files
app.py
CHANGED
@@ -1,10 +1,10 @@
|
|
|
|
1 |
import gradio as gr
|
2 |
-
import torch as th
|
3 |
|
4 |
from langchain.document_loaders import PDFMinerLoader,CSVLoader ,UnstructuredWordDocumentLoader,TextLoader
|
5 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
6 |
from langchain.embeddings import SentenceTransformerEmbeddings
|
7 |
-
from langchain.vectorstores import
|
8 |
from langchain import HuggingFaceHub
|
9 |
|
10 |
|
@@ -16,6 +16,15 @@ def loading_pdf():
|
|
16 |
return "Loading..."
|
17 |
|
18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
def process_documents(documents,data_chunk=1000,chunk_overlap=50):
|
20 |
text_splitter = RecursiveCharacterTextSplitter(chunk_size=data_chunk, chunk_overlap=chunk_overlap)
|
21 |
texts = text_splitter.split_documents(documents[0])
|
@@ -27,8 +36,7 @@ def get_hugging_face_model(model_id,API_key,temperature=0.1):
|
|
27 |
model_kwargs={"temperature": temperature, "max_new_tokens": 2048})
|
28 |
return chat_llm
|
29 |
|
30 |
-
def
|
31 |
-
|
32 |
embedding_model = SentenceTransformerEmbeddings(model_name='all-mpnet-base-v2',model_kwargs={"device": DEVICE})
|
33 |
|
34 |
document = None
|
@@ -43,6 +51,12 @@ def document_loading(file_data,doc_type='pdf',key=None):
|
|
43 |
|
44 |
texts = process_documents(documents=document)
|
45 |
vectordb = FAISS.from_documents(documents=texts, embedding= embedding_model)
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
|
47 |
|
48 |
def process_text_document(document_file_name):
|
@@ -104,3 +118,5 @@ with gr.Blocks(css=css) as demo:
|
|
104 |
chatbot = gr.Chatbot()
|
105 |
question = gr.Textbox(label="Question", placeholder="Type your question and hit Enter")
|
106 |
submit_button = gr.Button("Send Message")
|
|
|
|
|
|
1 |
+
import os
|
2 |
import gradio as gr
|
|
|
3 |
|
4 |
from langchain.document_loaders import PDFMinerLoader,CSVLoader ,UnstructuredWordDocumentLoader,TextLoader
|
5 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
6 |
from langchain.embeddings import SentenceTransformerEmbeddings
|
7 |
+
from langchain.vectorstores import FAISS
|
8 |
from langchain import HuggingFaceHub
|
9 |
|
10 |
|
|
|
16 |
return "Loading..."
|
17 |
|
18 |
|
19 |
+
def get_openai_chat_model(API_key):
|
20 |
+
try:
|
21 |
+
from langchain.llms import OpenAI
|
22 |
+
except ImportError as err:
|
23 |
+
raise "{}, unable to load openAI. Please install openai and add OPENAIAPI_KEY"
|
24 |
+
os.environ["OPENAI_API_KEY"] = API_key
|
25 |
+
llm = OpenAI()
|
26 |
+
return llm
|
27 |
+
|
28 |
def process_documents(documents,data_chunk=1000,chunk_overlap=50):
|
29 |
text_splitter = RecursiveCharacterTextSplitter(chunk_size=data_chunk, chunk_overlap=chunk_overlap)
|
30 |
texts = text_splitter.split_documents(documents[0])
|
|
|
36 |
model_kwargs={"temperature": temperature, "max_new_tokens": 2048})
|
37 |
return chat_llm
|
38 |
|
39 |
+
def chat_api(file_data,doc_type='pdf',key=None,llm_model='HuggingFace'):
|
|
|
40 |
embedding_model = SentenceTransformerEmbeddings(model_name='all-mpnet-base-v2',model_kwargs={"device": DEVICE})
|
41 |
|
42 |
document = None
|
|
|
51 |
|
52 |
texts = process_documents(documents=document)
|
53 |
vectordb = FAISS.from_documents(documents=texts, embedding= embedding_model)
|
54 |
+
if llm_model == 'HuggingFace':
|
55 |
+
llm = get_hugging_face_model(model_id='tiiuae/falcon-7b-instruct',API_key=key)
|
56 |
+
else:
|
57 |
+
llm_model = get_openai_chat_model(API_key=key)
|
58 |
+
|
59 |
+
|
60 |
|
61 |
|
62 |
def process_text_document(document_file_name):
|
|
|
118 |
chatbot = gr.Chatbot()
|
119 |
question = gr.Textbox(label="Question", placeholder="Type your question and hit Enter")
|
120 |
submit_button = gr.Button("Send Message")
|
121 |
+
load_pdf.click(loading_pdf, None, langchain_status, queue=False)
|
122 |
+
load_pdf.click(chat_api, inputs=[pdf_doc,file_extension,API_key,LLM_option], outputs=[langchain_status], queue=False)
|