Spaces:
Sleeping
Sleeping
initial commit
Browse files- app.py +75 -1
- dataset.csv +0 -0
app.py
CHANGED
@@ -1 +1,75 @@
|
|
1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import textwrap
|
3 |
+
from dotenv import load_dotenv
|
4 |
+
import gradio as gr
|
5 |
+
from haystack import Pipeline
|
6 |
+
from haystack.utils import Secret
|
7 |
+
from haystack.components.builders import PromptBuilder
|
8 |
+
from haystack.components.generators import OpenAIGenerator
|
9 |
+
import pandas as pd
|
10 |
+
|
11 |
+
load_dotenv()
|
12 |
+
|
13 |
+
MODEL = "microsoft/Phi-3-mini-4k-instruct"
|
14 |
+
|
15 |
+
# Load the CSV file
|
16 |
+
df = pd.read_csv("dataset.csv")
|
17 |
+
|
18 |
+
# Set up components
|
19 |
+
prompt_template = """
|
20 |
+
Based on the Indian Union Budget data for FY 21-22 to 23-24:
|
21 |
+
|
22 |
+
{{budget_data}}
|
23 |
+
|
24 |
+
Answer the given question: {{query}}
|
25 |
+
Answer:
|
26 |
+
"""
|
27 |
+
prompt_builder = PromptBuilder(template=prompt_template)
|
28 |
+
llm = OpenAIGenerator(
|
29 |
+
api_key=Secret.from_env_var("MONSTER_API_KEY"),
|
30 |
+
api_base_url="https://llm.monsterapi.ai/v1/",
|
31 |
+
model=MODEL,
|
32 |
+
generation_kwargs={"max_tokens": 512}
|
33 |
+
)
|
34 |
+
pipeline = Pipeline()
|
35 |
+
pipeline.add_component("prompt", prompt_builder)
|
36 |
+
pipeline.add_component("llm", llm)
|
37 |
+
|
38 |
+
pipeline.connect("prompt.prompt", "llm.prompt")
|
39 |
+
|
40 |
+
# Function to handle the query
|
41 |
+
# def answer_query(query):
|
42 |
+
# # Convert DataFrame to string representation
|
43 |
+
# budget_data = df.to_string()
|
44 |
+
# result = pipeline.run({"prompt": {"budget_data": budget_data, "query": query}})
|
45 |
+
# return result["llm"]["replies"][0]
|
46 |
+
|
47 |
+
|
48 |
+
def answer_query(query):
|
49 |
+
try:
|
50 |
+
# Select a subset of the data (adjust as needed)
|
51 |
+
sample_data = df.sample(n=10).to_string()
|
52 |
+
|
53 |
+
# Truncate the data if it's too long
|
54 |
+
budget_data = textwrap.shorten(sample_data, width=1000, placeholder="...")
|
55 |
+
|
56 |
+
result = pipeline.run({"prompt": {"budget_data": budget_data, "query": query}})
|
57 |
+
return result["llm"]["replies"][0]
|
58 |
+
except Exception as e:
|
59 |
+
return f"An error occurred: {str(e)}"
|
60 |
+
|
61 |
+
# Gradio interface
|
62 |
+
def chat_interface(query):
|
63 |
+
return answer_query(query)
|
64 |
+
|
65 |
+
with gr.Blocks() as demo:
|
66 |
+
gr.Markdown("# Indian 2024 Budget Chatbot")
|
67 |
+
query_input = gr.Textbox(label="Enter Your Question")
|
68 |
+
submit_button = gr.Button("Get Answer")
|
69 |
+
output_text = gr.Textbox(label="Answer", interactive=False)
|
70 |
+
|
71 |
+
submit_button.click(fn=chat_interface, inputs=[query_input], outputs=output_text)
|
72 |
+
|
73 |
+
# Run the app locally
|
74 |
+
if __name__ == "__main__":
|
75 |
+
demo.launch()
|
dataset.csv
ADDED
The diff for this file is too large to render.
See raw diff
|
|