Use local inference.
Browse files- app.py +9 -4
- requirements.txt +1 -0
app.py
CHANGED
@@ -1,24 +1,29 @@
|
|
1 |
import gradio as gr
|
2 |
from langchain import HuggingFaceHub
|
3 |
from langchain.chains.question_answering import load_qa_chain
|
4 |
-
from langchain.document_loaders import
|
5 |
from langchain.embeddings import HuggingFaceEmbeddings
|
|
|
6 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
7 |
from langchain.vectorstores import FAISS
|
8 |
-
from timeit import default_timer as timer
|
9 |
|
10 |
|
|
|
11 |
loader = TextLoader("rdna3.txt")
|
12 |
documents = loader.load()
|
13 |
|
|
|
14 |
splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=50)
|
15 |
chunks = splitter.split_documents(documents)
|
16 |
|
|
|
17 |
embeddings = HuggingFaceEmbeddings()
|
18 |
db = FAISS.from_documents(chunks, embeddings)
|
19 |
|
20 |
-
|
21 |
-
|
|
|
|
|
22 |
model_kwargs={"temperature": 0, "max_length": 128})
|
23 |
chain = load_qa_chain(llm, chain_type="stuff")
|
24 |
|
|
|
1 |
import gradio as gr
|
2 |
from langchain import HuggingFaceHub
|
3 |
from langchain.chains.question_answering import load_qa_chain
|
4 |
+
from langchain.document_loaders import TextLoader
|
5 |
from langchain.embeddings import HuggingFaceEmbeddings
|
6 |
+
from langchain.llms import HuggingFacePipeline
|
7 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
8 |
from langchain.vectorstores import FAISS
|
|
|
9 |
|
10 |
|
11 |
+
print("Loading documents")
|
12 |
loader = TextLoader("rdna3.txt")
|
13 |
documents = loader.load()
|
14 |
|
15 |
+
print("Creating chunks")
|
16 |
splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=50)
|
17 |
chunks = splitter.split_documents(documents)
|
18 |
|
19 |
+
print("Creating database")
|
20 |
embeddings = HuggingFaceEmbeddings()
|
21 |
db = FAISS.from_documents(chunks, embeddings)
|
22 |
|
23 |
+
print("Loading model")
|
24 |
+
llm = HuggingFacePipeline.from_model_id(
|
25 |
+
model_id="google/flan-t5-base",
|
26 |
+
task="text2text-generation",
|
27 |
model_kwargs={"temperature": 0, "max_length": 128})
|
28 |
chain = load_qa_chain(llm, chain_type="stuff")
|
29 |
|
requirements.txt
CHANGED
@@ -1,3 +1,4 @@
|
|
1 |
langchain
|
2 |
faiss-cpu
|
3 |
sentence_transformers
|
|
|
|
1 |
langchain
|
2 |
faiss-cpu
|
3 |
sentence_transformers
|
4 |
+
protobuf==3.20.1
|