Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,22 +1,58 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
-
from
|
|
|
|
|
|
|
| 3 |
|
| 4 |
-
#
|
|
|
|
|
|
|
| 5 |
model, tokenizer, device = load_and_train()
|
| 6 |
|
| 7 |
-
#
|
| 8 |
-
|
| 9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
iface = gr.Interface(
|
| 12 |
-
fn=
|
| 13 |
inputs=[
|
| 14 |
gr.Textbox(label="Enter your question:", lines=5, placeholder="Type your question here..."),
|
| 15 |
gr.Slider(minimum=50, maximum=500, step=10, value=200, label="Max tokens to generate")
|
| 16 |
],
|
| 17 |
-
outputs=
|
| 18 |
-
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
| 20 |
).queue()
|
| 21 |
|
| 22 |
iface.launch()
|
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
+
from concurrent.futures import ThreadPoolExecutor
|
| 3 |
+
import fine_tuned
|
| 4 |
+
from fine_tuned import load_and_train
|
| 5 |
|
| 6 |
+
# -----------------------------
|
| 7 |
+
# Load fine-tuned model
|
| 8 |
+
# -----------------------------
|
| 9 |
model, tokenizer, device = load_and_train()
|
| 10 |
|
| 11 |
+
# -----------------------------
|
| 12 |
+
# RAG Backend (Stub Example)
|
| 13 |
+
# -----------------------------
|
| 14 |
+
def generate_answer_rag(prompt, max_tokens=200):
|
| 15 |
+
"""
|
| 16 |
+
Replace this stub with your actual RAG pipeline.
|
| 17 |
+
For now, just a dummy response.
|
| 18 |
+
"""
|
| 19 |
+
return f"[RAG answer for]: {prompt[:50]}..."
|
| 20 |
|
| 21 |
+
|
| 22 |
+
# -----------------------------
|
| 23 |
+
# Combined Answer Function
|
| 24 |
+
# -----------------------------
|
| 25 |
+
def combined_generate(prompt, max_tokens):
|
| 26 |
+
with ThreadPoolExecutor() as executor:
|
| 27 |
+
# Submit both tasks in parallel
|
| 28 |
+
ft_future = executor.submit(fine_tuned.generate_answer, model, tokenizer, device, prompt, max_tokens)
|
| 29 |
+
rag_future = executor.submit(generate_answer_rag, prompt, max_tokens)
|
| 30 |
+
|
| 31 |
+
fine_tuned_answer = ft_future.result()
|
| 32 |
+
rag_answer = rag_future.result()
|
| 33 |
+
|
| 34 |
+
return {
|
| 35 |
+
"Fine-tuned Model Answer": fine_tuned_answer,
|
| 36 |
+
"RAG Answer": rag_answer
|
| 37 |
+
}
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
# -----------------------------
|
| 41 |
+
# Gradio Interface
|
| 42 |
+
# -----------------------------
|
| 43 |
iface = gr.Interface(
|
| 44 |
+
fn=combined_generate,
|
| 45 |
inputs=[
|
| 46 |
gr.Textbox(label="Enter your question:", lines=5, placeholder="Type your question here..."),
|
| 47 |
gr.Slider(minimum=50, maximum=500, step=10, value=200, label="Max tokens to generate")
|
| 48 |
],
|
| 49 |
+
outputs=[
|
| 50 |
+
gr.Textbox(label="Fine-tuned Model Answer"),
|
| 51 |
+
gr.Textbox(label="RAG Answer")
|
| 52 |
+
],
|
| 53 |
+
title="Compare Fine-tuned Model vs RAG 🤖📚",
|
| 54 |
+
description="Ask a question and get answers from both the fine-tuned model and the RAG pipeline."
|
| 55 |
).queue()
|
| 56 |
|
| 57 |
iface.launch()
|
| 58 |
+
|