Update app.py
Browse files
app.py
CHANGED
|
@@ -1,5 +1,28 @@
|
|
| 1 |
-
|
| 2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
from langchain_community.embeddings import HuggingFaceEmbeddings
|
| 4 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
gr.load("models/HuggingFaceH4/zephyr-7b-beta").launch()
|
|
|
|
| 1 |
+
from langchain.document_loaders import PyPDFLoader
|
| 2 |
+
loader = PyPDFLoader("Test/blob/main/Footcare.pdf")
|
| 3 |
+
pages = loader.load()
|
| 4 |
+
|
| 5 |
+
|
| 6 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
| 7 |
+
splitter = RecursiveCharacterTextSplitter(chunk_size=512, chunk_overlap=30)
|
| 8 |
+
docs = splitter.split_documents(pages)
|
| 9 |
+
|
| 10 |
+
|
| 11 |
from langchain_community.embeddings import HuggingFaceEmbeddings
|
| 12 |
+
embedding_model = HuggingFaceEmbedding(model_name="all-MiniLM-L6-v2")
|
| 13 |
+
embeddings = model.encode(docs)
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
from langchain.vectorstores import Chroma
|
| 17 |
+
persist_directory = 'docs/chroma/'
|
| 18 |
+
vectordb = Chroma.from_documents(
|
| 19 |
+
documents=docs,
|
| 20 |
+
embedding=embedding,
|
| 21 |
+
persist_directory=persist_directory
|
| 22 |
+
)
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
retriever = vectordb.as_retriever()
|
| 26 |
+
import gradio as gr
|
| 27 |
+
import gradio as gr
|
| 28 |
gr.load("models/HuggingFaceH4/zephyr-7b-beta").launch()
|