Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,35 +1,36 @@
|
|
1 |
-
# app.py
|
2 |
import gradio as gr
|
3 |
import torch
|
4 |
import pandas as pd
|
5 |
-
from transformers import RagTokenizer, RagRetriever, RagSequenceForGeneration
|
6 |
import yfinance as yf
|
|
|
|
|
|
|
|
|
|
|
7 |
|
8 |
# Function to fetch and preprocess ICICI Bank data
|
9 |
def fetch_and_preprocess_data():
|
10 |
try:
|
11 |
-
#
|
12 |
-
ticker = "ICICIBANK.BO" # Use BSE symbol
|
13 |
data = yf.download(ticker, start="2020-01-01", end="2023-01-01")
|
14 |
|
15 |
if data.empty:
|
16 |
raise ValueError("No data found for the given symbol.")
|
17 |
|
18 |
-
# Calculate
|
19 |
data['MA_50'] = data['Close'].rolling(window=50).mean()
|
20 |
data['MA_200'] = data['Close'].rolling(window=200).mean()
|
21 |
|
22 |
return data
|
23 |
except Exception as e:
|
24 |
print(f"Error fetching data: {e}")
|
25 |
-
return pd.DataFrame() # Return
|
26 |
|
27 |
-
# Load the
|
28 |
try:
|
29 |
tokenizer = RagTokenizer.from_pretrained("facebook/rag-sequence-base")
|
30 |
print("Tokenizer loaded successfully.")
|
31 |
|
32 |
-
# Use a pre-built index (e.g., wiki_dpr) instead of creating a custom index
|
33 |
retriever = RagRetriever.from_pretrained(
|
34 |
"facebook/rag-sequence-base",
|
35 |
index_name="wiki_dpr",
|
@@ -38,45 +39,44 @@ try:
|
|
38 |
)
|
39 |
print("Retriever loaded successfully.")
|
40 |
|
41 |
-
|
42 |
-
model = RagSequenceForGeneration.from_pretrained("facebook/rag-sequence-base", retriever=retriever)
|
43 |
print("Model loaded successfully.")
|
44 |
except Exception as e:
|
45 |
-
print(f"Error
|
46 |
retriever = None
|
47 |
model = None
|
48 |
|
49 |
-
# Function to analyze trading data
|
50 |
def analyze_trading_data(question):
|
51 |
if model is None or retriever is None:
|
52 |
return "Error: Model or retriever is not initialized. Please check the logs."
|
53 |
-
|
54 |
# Fetch and preprocess data
|
55 |
data = fetch_and_preprocess_data()
|
56 |
-
|
57 |
if data.empty:
|
58 |
return "Error: No data available for analysis."
|
59 |
-
|
60 |
-
# Prepare context for
|
61 |
context = (
|
62 |
f"ICICI Bank stock data:\n"
|
63 |
f"Latest Close Price: {data['Close'].iloc[-1]:.2f}\n"
|
64 |
f"50-Day Moving Average: {data['MA_50'].iloc[-1]:.2f}\n"
|
65 |
f"200-Day Moving Average: {data['MA_200'].iloc[-1]:.2f}\n"
|
66 |
)
|
67 |
-
|
68 |
# Combine question and context
|
69 |
input_text = f"Question: {question}\nContext: {context}"
|
70 |
-
|
71 |
-
# Tokenize
|
72 |
-
inputs = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True)
|
73 |
-
|
74 |
-
# Generate
|
75 |
outputs = model.generate(inputs['input_ids'])
|
76 |
|
77 |
-
# Decode
|
78 |
answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
79 |
-
|
80 |
return answer
|
81 |
|
82 |
# Gradio interface
|
@@ -94,4 +94,5 @@ iface = gr.Interface(
|
|
94 |
)
|
95 |
|
96 |
# Launch the app
|
97 |
-
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
import pandas as pd
|
|
|
4 |
import yfinance as yf
|
5 |
+
from transformers import RagTokenizer, RagRetriever, RagSequenceForGeneration
|
6 |
+
|
7 |
+
# Check if GPU is available
|
8 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
9 |
+
print(f"Using device: {device}")
|
10 |
|
11 |
# Function to fetch and preprocess ICICI Bank data
|
12 |
def fetch_and_preprocess_data():
|
13 |
try:
|
14 |
+
ticker = "ICICIBANK.BO" # ICICI Bank BSE Symbol
|
|
|
15 |
data = yf.download(ticker, start="2020-01-01", end="2023-01-01")
|
16 |
|
17 |
if data.empty:
|
18 |
raise ValueError("No data found for the given symbol.")
|
19 |
|
20 |
+
# Calculate Moving Averages
|
21 |
data['MA_50'] = data['Close'].rolling(window=50).mean()
|
22 |
data['MA_200'] = data['Close'].rolling(window=200).mean()
|
23 |
|
24 |
return data
|
25 |
except Exception as e:
|
26 |
print(f"Error fetching data: {e}")
|
27 |
+
return pd.DataFrame() # Return empty DataFrame if fetching fails
|
28 |
|
29 |
+
# Load the RAG model and tokenizer with error handling
|
30 |
try:
|
31 |
tokenizer = RagTokenizer.from_pretrained("facebook/rag-sequence-base")
|
32 |
print("Tokenizer loaded successfully.")
|
33 |
|
|
|
34 |
retriever = RagRetriever.from_pretrained(
|
35 |
"facebook/rag-sequence-base",
|
36 |
index_name="wiki_dpr",
|
|
|
39 |
)
|
40 |
print("Retriever loaded successfully.")
|
41 |
|
42 |
+
model = RagSequenceForGeneration.from_pretrained("facebook/rag-sequence-base", retriever=retriever).to(device)
|
|
|
43 |
print("Model loaded successfully.")
|
44 |
except Exception as e:
|
45 |
+
print(f"Error loading model or retriever: {e}")
|
46 |
retriever = None
|
47 |
model = None
|
48 |
|
49 |
+
# Function to analyze trading data
|
50 |
def analyze_trading_data(question):
|
51 |
if model is None or retriever is None:
|
52 |
return "Error: Model or retriever is not initialized. Please check the logs."
|
53 |
+
|
54 |
# Fetch and preprocess data
|
55 |
data = fetch_and_preprocess_data()
|
56 |
+
|
57 |
if data.empty:
|
58 |
return "Error: No data available for analysis."
|
59 |
+
|
60 |
+
# Prepare context for RAG model
|
61 |
context = (
|
62 |
f"ICICI Bank stock data:\n"
|
63 |
f"Latest Close Price: {data['Close'].iloc[-1]:.2f}\n"
|
64 |
f"50-Day Moving Average: {data['MA_50'].iloc[-1]:.2f}\n"
|
65 |
f"200-Day Moving Average: {data['MA_200'].iloc[-1]:.2f}\n"
|
66 |
)
|
67 |
+
|
68 |
# Combine question and context
|
69 |
input_text = f"Question: {question}\nContext: {context}"
|
70 |
+
|
71 |
+
# Tokenize input
|
72 |
+
inputs = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True).to(device)
|
73 |
+
|
74 |
+
# Generate answer using the model
|
75 |
outputs = model.generate(inputs['input_ids'])
|
76 |
|
77 |
+
# Decode output
|
78 |
answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
79 |
+
|
80 |
return answer
|
81 |
|
82 |
# Gradio interface
|
|
|
94 |
)
|
95 |
|
96 |
# Launch the app
|
97 |
+
if __name__ == "__main__":
|
98 |
+
iface.launch()
|