Spaces:
Sleeping
Sleeping
[Winsmano Ambani] commited on
Commit ·
650f31e
1
Parent(s): 3568bf1
Deploy QA app
Browse files- .gitattributes +1 -0
- QA_model/config.json +3 -0
- QA_model/model.safetensors +3 -0
- QA_model/special_tokens_map.json +3 -0
- QA_model/tokenizer.json +3 -0
- QA_model/tokenizer_config.json +3 -0
- QA_model/training_args.bin +3 -0
- QA_model/vocab.txt +3 -0
- app.py +51 -0
- data/SalatusSalatin.txt +0 -0
- requirements.txt +5 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
QA_model/* filter=lfs diff=lfs merge=lfs -text
|
QA_model/config.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:74043a90e859afee56cf6d011a1b0be1a4a7fd75fe16f519aeb10e49ded58db3
|
| 3 |
+
size 790
|
QA_model/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f62e576b06351a51efc17849cf331c64bf077f5c5de274dbf905c6365782b643
|
| 3 |
+
size 709080904
|
QA_model/special_tokens_map.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b6d346be366a7d1d48332dbc9fdf3bf8960b5d879522b7799ddba59e76237ee3
|
| 3 |
+
size 125
|
QA_model/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18dddf49b787531e9c57309e6a09fe6901cda3784fa0b9ddbeaa3169fd0cca75
|
| 3 |
+
size 2919627
|
QA_model/tokenizer_config.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c441dfe412d9d7e47c960029a48a0159c23a38a9ab41b465d90fb1f520d4ced
|
| 3 |
+
size 1222
|
QA_model/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3096b7c0a7c212d09a94e1e9770b3cccc615d56457f4540620442d0e12cf0578
|
| 3 |
+
size 5240
|
QA_model/vocab.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fe0fda7c425b48c516fc8f160d594c8022a0808447475c1a7c6d6479763f310c
|
| 3 |
+
size 995526
|
app.py
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import faiss
|
| 3 |
+
import torch
|
| 4 |
+
from transformers import AutoTokenizer, AutoModelForQuestionAnswering, pipeline
|
| 5 |
+
from sentence_transformers import SentenceTransformer
|
| 6 |
+
import os
|
| 7 |
+
|
| 8 |
+
# Load QA model
|
| 9 |
+
qa_model_name = "Deployment\SampleTest\QA_model"
|
| 10 |
+
qa_tokenizer = AutoTokenizer.from_pretrained(qa_model_name)
|
| 11 |
+
qa_model = AutoModelForQuestionAnswering.from_pretrained(qa_model_name)
|
| 12 |
+
|
| 13 |
+
# Load embedding model for retrieval
|
| 14 |
+
embedder = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
|
| 15 |
+
|
| 16 |
+
# Load knowledge base
|
| 17 |
+
with open("Deployment\SampleTest\data\SalatusSalatin.txt", "r", encoding="utf-8") as f:
|
| 18 |
+
docs = f.read().split("\n\n") # split into paragraphs
|
| 19 |
+
|
| 20 |
+
# Create FAISS index
|
| 21 |
+
embeddings = embedder.encode(docs, convert_to_numpy=True, show_progress_bar=True)
|
| 22 |
+
dimension = embeddings.shape[1]
|
| 23 |
+
index = faiss.IndexFlatL2(dimension)
|
| 24 |
+
index.add(embeddings)
|
| 25 |
+
|
| 26 |
+
def answer_question(question):
|
| 27 |
+
# Step 1: retrieve top context
|
| 28 |
+
q_emb = embedder.encode([question], convert_to_numpy=True)
|
| 29 |
+
D, I = index.search(q_emb, k=1) # top-1 context
|
| 30 |
+
context = docs[I[0][0]]
|
| 31 |
+
|
| 32 |
+
# Step 2: QA extraction
|
| 33 |
+
inputs = qa_tokenizer(question, context, return_tensors="pt")
|
| 34 |
+
outputs = qa_model(**inputs)
|
| 35 |
+
|
| 36 |
+
start = torch.argmax(outputs.start_logits)
|
| 37 |
+
end = torch.argmax(outputs.end_logits)
|
| 38 |
+
answer = qa_tokenizer.decode(inputs["input_ids"][0][start:end+1])
|
| 39 |
+
return f"Answer: {answer}\n\n(Context: {context})"
|
| 40 |
+
|
| 41 |
+
# Gradio UI
|
| 42 |
+
iface = gr.Interface(
|
| 43 |
+
fn=answer_question,
|
| 44 |
+
inputs="text",
|
| 45 |
+
outputs="text",
|
| 46 |
+
title="Retrieval-Augmented QA",
|
| 47 |
+
description="Ask a question, the system retrieves relevant context from a large text file and extracts the answer."
|
| 48 |
+
)
|
| 49 |
+
|
| 50 |
+
if __name__ == "__main__":
|
| 51 |
+
iface.launch()
|
data/SalatusSalatin.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
requirements.txt
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
transformers
|
| 2 |
+
torch
|
| 3 |
+
gradio
|
| 4 |
+
faiss-cpu
|
| 5 |
+
sentence-transformers
|