Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -4,6 +4,7 @@ from tempfile import NamedTemporaryFile
|
|
4 |
from langchain.document_loaders import PyPDFLoader
|
5 |
from langchain.embeddings import HuggingFaceEmbeddings
|
6 |
from langchain.vectorstores import Chroma
|
|
|
7 |
from transformers import AutoModelForQuestionAnswering, AutoTokenizer, pipeline
|
8 |
|
9 |
# Function to save the uploaded PDF to a temporary file
|
@@ -27,19 +28,14 @@ if uploaded_file is not None:
|
|
27 |
embed = HuggingFaceEmbeddings()
|
28 |
db = Chroma.from_documents(pages, embed)
|
29 |
|
30 |
-
# Initialize the GPT-4 model
|
31 |
-
llm = GPT4All(
|
32 |
-
model="./gpt4all-converted.bin"
|
33 |
-
)
|
34 |
-
|
35 |
# Define a function to get answers
|
36 |
def get_answer(question):
|
37 |
doc = db.similarity_search(question, k=4)
|
38 |
context = doc[0].page_content + doc[1].page_content + doc[2].page_content + doc[3].page_content
|
39 |
|
40 |
-
max_seq_length = 512 # You may define this based on your model
|
41 |
|
42 |
-
context = context[:max_seq_length]
|
43 |
|
44 |
# Load the model & tokenizer for question-answering
|
45 |
model_name = "deepset/roberta-base-squad2"
|
|
|
4 |
from langchain.document_loaders import PyPDFLoader
|
5 |
from langchain.embeddings import HuggingFaceEmbeddings
|
6 |
from langchain.vectorstores import Chroma
|
7 |
+
from langchain import PromptTemplate, LLMChain
|
8 |
from transformers import AutoModelForQuestionAnswering, AutoTokenizer, pipeline
|
9 |
|
10 |
# Function to save the uploaded PDF to a temporary file
|
|
|
28 |
embed = HuggingFaceEmbeddings()
|
29 |
db = Chroma.from_documents(pages, embed)
|
30 |
|
|
|
|
|
|
|
|
|
|
|
31 |
# Define a function to get answers
|
32 |
def get_answer(question):
|
33 |
doc = db.similarity_search(question, k=4)
|
34 |
context = doc[0].page_content + doc[1].page_content + doc[2].page_content + doc[3].page_content
|
35 |
|
36 |
+
#max_seq_length = 512 # You may define this based on your model
|
37 |
|
38 |
+
#context = context[:max_seq_length]
|
39 |
|
40 |
# Load the model & tokenizer for question-answering
|
41 |
model_name = "deepset/roberta-base-squad2"
|