Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,104 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pandas as pd
|
2 |
+
import numpy as np
|
3 |
+
from sklearn.feature_extraction.text import TfidfVectorizer
|
4 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
5 |
+
from transformers import pipeline, HfApi, HfFolder
|
6 |
+
import gradio as gr
|
7 |
+
|
8 |
+
# Hugging Face login - Ensure that you are logged into Hugging Face before running this space
|
9 |
+
HfFolder.save_token("<your_hugging_face_token>") # This is set once to avoid token exposure in the code
|
10 |
+
|
11 |
+
# Load and preprocess the data
|
12 |
+
def preprocess_data(file_path):
|
13 |
+
"""Load and preprocess the CSV data."""
|
14 |
+
data = pd.read_csv(file_path)
|
15 |
+
|
16 |
+
# Clean column names
|
17 |
+
data.columns = data.columns.str.strip().str.replace('#', 'Count').str.replace(' ', '_')
|
18 |
+
|
19 |
+
# Handle missing values (if any)
|
20 |
+
data.fillna(0, inplace=True)
|
21 |
+
|
22 |
+
return data
|
23 |
+
|
24 |
+
# Convert data into a retrievable knowledge base
|
25 |
+
def create_knowledge_base(data):
|
26 |
+
"""Transform the data into a knowledge base suitable for retrieval."""
|
27 |
+
# Combine relevant fields into a single text-based feature for embedding
|
28 |
+
data['Knowledge_Text'] = data.apply(lambda row: (
|
29 |
+
f"Player: {row['Player_Name']}, Position: {row['Main_Possition']}, "
|
30 |
+
f"Date: {row['Date']}, Session: {row['Session_Name']}, "
|
31 |
+
f"Played Time: {row['Played_Time_(min)']} minutes, Top Speed: {row['Top_Speed_(km/h)']} km/h, "
|
32 |
+
f"Distance Covered: {row['Dist._Covered_(m)']} meters, "
|
33 |
+
f"Intensity: {row['Session_Intensity']}, "
|
34 |
+
f"RPE: {row['RPE']}, s-RPE: {row['s-RPE']}"
|
35 |
+
), axis=1)
|
36 |
+
|
37 |
+
return data[['Player_ID', 'Knowledge_Text']]
|
38 |
+
|
39 |
+
# Create a similarity-based retrieval function
|
40 |
+
def query_knowledge_base(knowledge_base, query, vectorizer):
|
41 |
+
"""Query the knowledge base using cosine similarity."""
|
42 |
+
query_vec = vectorizer.transform([query])
|
43 |
+
knowledge_vec = vectorizer.transform(knowledge_base['Knowledge_Text'])
|
44 |
+
|
45 |
+
# Compute cosine similarities
|
46 |
+
similarities = cosine_similarity(query_vec, knowledge_vec).flatten()
|
47 |
+
|
48 |
+
# Retrieve the most relevant rows
|
49 |
+
top_indices = np.argsort(similarities)[::-1][:5] # Top 5 results
|
50 |
+
return knowledge_base.iloc[top_indices], similarities[top_indices]
|
51 |
+
|
52 |
+
# Main pipeline with LLM integration and prompt engineering
|
53 |
+
def main_pipeline(file_path, user_query):
|
54 |
+
"""End-to-end pipeline for the RAG system with Llama3.2 and prompt engineering."""
|
55 |
+
# Preprocess data
|
56 |
+
data = preprocess_data(file_path)
|
57 |
+
knowledge_base = create_knowledge_base(data)
|
58 |
+
|
59 |
+
# Create TF-IDF Vectorizer
|
60 |
+
vectorizer = TfidfVectorizer()
|
61 |
+
vectorizer.fit(knowledge_base['Knowledge_Text'])
|
62 |
+
|
63 |
+
# Query the knowledge base
|
64 |
+
results, scores = query_knowledge_base(knowledge_base, user_query, vectorizer)
|
65 |
+
|
66 |
+
# Format retrieved knowledge for LLM input
|
67 |
+
retrieved_text = "\n".join(results['Knowledge_Text'].tolist())
|
68 |
+
|
69 |
+
# Use Llama3.2 for question answering with prompt engineering
|
70 |
+
llm = pipeline("text-generation", model="meta-llama/Llama-3.2-1B-Instruct")
|
71 |
+
prompt = (
|
72 |
+
f"You are an expert sports analyst. Based on the following training data, provide a detailed and insightful answer to the user's question. "
|
73 |
+
f"Always include relevant numerical data in your response. Limit your response to a maximum of 200 words.\n\n"
|
74 |
+
f"Training Data:\n{retrieved_text}\n\n"
|
75 |
+
f"User Question: {user_query}\n\nAnswer:"
|
76 |
+
)
|
77 |
+
response = llm(prompt, max_new_tokens=200, num_return_sequences=1)
|
78 |
+
|
79 |
+
# Extract the answer part only
|
80 |
+
answer = response[0]['generated_text'].split("Answer:", 1)[-1].strip()
|
81 |
+
return answer
|
82 |
+
|
83 |
+
# Gradio interface
|
84 |
+
def query_interface(file_path, user_query):
|
85 |
+
try:
|
86 |
+
result = main_pipeline(file_path.name, user_query)
|
87 |
+
return result
|
88 |
+
except Exception as e:
|
89 |
+
return str(e)
|
90 |
+
|
91 |
+
# Launch Gradio app
|
92 |
+
file_input = gr.File(label="Upload CSV File")
|
93 |
+
text_input = gr.Textbox(label="Ask a Question", lines=2, placeholder="Enter your query here...")
|
94 |
+
output = gr.Textbox(label="Answer")
|
95 |
+
|
96 |
+
interface = gr.Interface(
|
97 |
+
fn=query_interface,
|
98 |
+
inputs=[file_input, text_input],
|
99 |
+
outputs=output,
|
100 |
+
title="RAG Training Data Query System",
|
101 |
+
description="Upload a CSV file containing training data and ask detailed questions about it."
|
102 |
+
)
|
103 |
+
|
104 |
+
interface.launch()
|