Spaces:
Sleeping
Sleeping
yashasgupta
commited on
Commit
•
8f74b08
1
Parent(s):
bd8a766
Update app.py
Browse files
app.py
CHANGED
@@ -44,11 +44,10 @@ from langchain_text_splitters import NLTKTextSplitter
|
|
44 |
|
45 |
uploaded_file = st.file_uploader("Choose a pdf file",type = "pdf")
|
46 |
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
chunks = test_splitter.split_documents(dat_nik)
|
52 |
|
53 |
# dat = PDFMinerLoader("2404.07143.pdf")
|
54 |
# dat_nik =dat.load()
|
|
|
44 |
|
45 |
uploaded_file = st.file_uploader("Choose a pdf file",type = "pdf")
|
46 |
|
47 |
+
pdf_loader = PDFMinerLoader(uploaded_file)
|
48 |
+
dat_nik = pdf_loader.load()
|
49 |
+
text_splitter = NLTKTextSplitter(chunk_size = 500,chunk_overlap = 100)
|
50 |
+
chunks = test_splitter.split_documents(dat_nik)
|
|
|
51 |
|
52 |
# dat = PDFMinerLoader("2404.07143.pdf")
|
53 |
# dat_nik =dat.load()
|