Upload 8 files

FINGPT_MIGRATION.md

@@ -0,0 +1,92 @@
+# FinGPT Migration Guide
+
+## Overview
+This document describes the migration from Google Gemini 2.5 Pro to FinGPT model for the FinRobot Forecaster application.
+
+## Changes Made
+
+### 1. Dependencies Updated
+- **Removed**: `google-generativeai>=0.3.0`
+- **Added**: 
+  - `transformers>=4.30.0`
+  - `torch>=2.0.0`
+  - `accelerate>=0.20.0`
+  - `peft>=0.4.0`
+  - `bitsandbytes>=0.39.0`
+
+### 2. Model Configuration
+- **Old**: `GEMINI_MODEL = "gemini-2.5-pro"`
+- **New**: `FINGPT_MODEL_NAME = "Starfish55/fingpt-complete"`
+
+### 3. API Key Changes
+- **Old**: `GOOGLE_API_KEYS` environment variable
+- **New**: `HF_TOKEN` environment variable for Hugging Face access
+
+### 4. Model Loading
+- **Old**: Google Generative AI client initialization
+- **New**: Hugging Face Transformers with BitsAndBytesConfig for memory efficiency
+
+### 5. Inference Function
+- **Old**: `genai.GenerativeModel.generate_content()`
+- **New**: `model.generate()` with custom tokenization and decoding
+
+## Environment Variables Required
+
+### For Hugging Face Spaces:
+```bash
+HF_TOKEN=your_hugging_face_token_here
+FINNHUB_KEYS=your_finnhub_api_keys_here
+RAPIDAPI_KEYS=your_rapidapi_keys_here
+```
+
+## Model Features
+
+### FinGPT Advantages:
+1. **Specialized for Finance**: Trained specifically on financial data
+2. **Open Source**: No API rate limits or costs
+3. **Memory Efficient**: Uses 4-bit quantization
+4. **Real-time Updates**: Can be fine-tuned with latest financial data
+
+### Performance Considerations:
+- **Memory Usage**: ~4-8GB GPU memory (with quantization)
+- **Inference Speed**: Slower than API calls but more reliable
+- **Model Size**: ~7B parameters (much smaller than Gemini)
+
+## Testing
+
+Run the test script to verify the integration:
+```bash
+python test_fingpt_integration.py
+```
+
+## Deployment Notes
+
+1. **Hugging Face Spaces**: Ensure `HF_TOKEN` is set in secrets
+2. **GPU Requirements**: Requires GPU with at least 8GB VRAM
+3. **Model Loading**: First run may take longer due to model download
+4. **Fallback**: App will use mock responses if model fails to load
+
+## Troubleshooting
+
+### Common Issues:
+1. **Out of Memory**: Reduce `MAX_LENGTH` or use CPU inference
+2. **Model Loading Fails**: Check `HF_TOKEN` and internet connection
+3. **Slow Inference**: Consider using smaller model or CPU inference
+
+### Debug Mode:
+Set `debug=True` in the app launch to see detailed error messages.
+
+## Migration Benefits
+
+1. **Cost Effective**: No API costs for inference
+2. **Privacy**: Data stays local, no external API calls
+3. **Reliability**: No rate limits or API downtime
+4. **Customization**: Can fine-tune for specific financial tasks
+5. **Transparency**: Full control over model behavior
+
+## Next Steps
+
+1. Test the application with real financial data
+2. Fine-tune the model if needed for specific use cases
+3. Monitor performance and adjust parameters
+4. Consider implementing model caching for faster startup

UPLOAD_GUIDE.md

@@ -0,0 +1,103 @@
+# Hướng dẫn Upload lên Hugging Face Space
+
+## 📋 Các file cần thiết đã được tạo
+
+✅ **app.py** - File chính của ứng dụng  
+✅ **README.md** - Mô tả Space với metadata Hugging Face  
+✅ **requirements.txt** - Dependencies cần thiết  
+✅ **config.json** - Cấu hình Space  
+✅ **.gitignore** - Loại trừ file không cần thiết  
+
+## 🚀 Cách upload lên Hugging Face Space
+
+### Phương pháp 1: Sử dụng Git (Khuyến nghị)
+
+1. **Tạo Space mới trên Hugging Face:**
+   - Truy cập: https://huggingface.co/new-space
+   - Đặt tên: `your-username/finhigh` (thay your-username bằng tên của bạn)
+   - Chọn SDK: Gradio
+   - Chọn License: MIT
+   - Tạo Space
+
+2. **Clone Space về máy:**
+   ```bash
+   git clone https://huggingface.co/spaces/your-username/finhigh
+   cd finhigh
+   ```
+
+3. **Copy files vào Space:**
+   ```bash
+   # Copy tất cả files từ FinHigh_HuggingFace_Upload/
+   cp -r ../FinHigh_HuggingFace_Upload/* .
+   ```
+
+4. **Commit và push:**
+   ```bash
+   git add .
+   git commit -m "Initial commit: FinHigh Stock Prediction App"
+   git push
+   ```
+
+### Phương pháp 2: Upload trực tiếp
+
+1. **Tạo Space mới trên Hugging Face**
+2. **Upload từng file:**
+   - Upload `app.py` làm file chính
+   - Upload `requirements.txt`
+   - Upload `README.md`
+   - Upload `config.json` (nếu cần)
+
+## ⚙️ Thiết lập API Keys
+
+Sau khi upload, cần thiết lập các API keys trong Settings của Space:
+
+1. **Vào Settings của Space**
+2. **Thêm các Secrets:**
+   - `FINNHUB_KEYS`: API keys cho Finnhub (phân cách bằng dòng mới)
+   - `RAPIDAPI_KEYS`: API keys cho RapidAPI (phân cách bằng dòng mới)
+   - `GOOGLE_API_KEYS`: API keys cho Google Generative AI (phân cách bằng dòng mới)
+
+## 🔧 Cấu hình Space
+
+- **Hardware**: CPU (đã cấu hình trong config.json)
+- **Memory**: 2GB
+- **Disk**: 10GB
+- **SDK**: Gradio 4.44.0
+
+## 📊 Kiểm tra sau khi upload
+
+1. **Space sẽ tự động build** sau khi push code
+2. **Kiểm tra logs** nếu có lỗi
+3. **Test ứng dụng** với các ví dụ có sẵn
+4. **Kiểm tra API keys** hoạt động đúng
+
+## 🐛 Troubleshooting
+
+### Lỗi thường gặp:
+
+1. **Import Error**: Kiểm tra requirements.txt
+2. **API Key Error**: Kiểm tra Secrets trong Settings
+3. **Memory Error**: Tăng memory limit trong config.json
+4. **Timeout Error**: Tối ưu code hoặc tăng timeout
+
+### Giải pháp:
+
+- Xem logs trong tab "Logs" của Space
+- Kiểm tra file requirements.txt
+- Đảm bảo API keys đúng format
+- Test local trước khi upload
+
+## 📝 Lưu ý quan trọng
+
+- ✅ Code đã được tối ưu cho Hugging Face
+- ✅ Dependencies đã được pin version
+- ✅ README có metadata đúng format
+- ✅ Config.json đã cấu hình sẵn
+- ⚠️ Cần thiết lập API keys sau khi upload
+- ⚠️ Test kỹ trước khi public
+
+## 🔗 Liên kết hữu ích
+
+- [Hugging Face Spaces Documentation](https://huggingface.co/docs/hub/spaces)
+- [Gradio Documentation](https://gradio.app/docs/)
+- [API Keys Guide](https://huggingface.co/docs/hub/spaces-sdks-docker-advanced#secrets)

app.py

@@ -7,24 +7,27 @@ from datetime import date, datetime, timedelta
 import gradio as gr
 import pandas as pd
 import finnhub
-import google.generativeai as genai
 from io import StringIO
 import requests
 from requests.adapters import HTTPAdapter
 from urllib3.util.retry import Retry
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+from peft import PeftModel
 
-# Suppress Google Cloud warnings
-os.environ['GRPC_VERBOSITY'] = 'ERROR'
-os.environ['GRPC_TRACE'] = ''
-
-# Suppress other warnings
+# Suppress warnings
 import warnings
 warnings.filterwarnings('ignore', category=UserWarning)
 warnings.filterwarnings('ignore', category=FutureWarning)
+warnings.filterwarnings('ignore', category=DeprecationWarning)
 
 # ---------- CẤU HÌNH ---------------------------------------------------------
 
-GEMINI_MODEL = "gemini-2.5-pro"
+# FinGPT Model Configuration
+FINGPT_MODEL_NAME = "Starfish55/fingpt-complete"
+FINGPT_BASE_MODEL = "microsoft/DialoGPT-medium"  # Base model for FinGPT
+MAX_LENGTH = 1024
+TEMPERATURE = 0.7
 
 # RapidAPI Configuration
 RAPIDAPI_HOST = "alpha-vantage.p.rapidapi.com"
@@ -43,27 +46,21 @@ if RAPIDAPI_KEYS_RAW:
 else:
     RAPIDAPI_KEYS = []
 
-# Load Google API keys from single secret (multiple keys separated by newlines)
-GOOGLE_API_KEYS_RAW = os.getenv("GOOGLE_API_KEYS", "")
-if GOOGLE_API_KEYS_RAW:
-    GOOGLE_API_KEYS = [key.strip() for key in GOOGLE_API_KEYS_RAW.split('\n') if key.strip()]
-else:
-    GOOGLE_API_KEYS = []
+# Load Hugging Face API token for FinGPT model access
+HF_TOKEN = os.getenv("HF_TOKEN", "")
+if not HF_TOKEN:
+    print("⚠️ Warning: No Hugging Face token found in secrets")
 
 # Filter out empty keys
 FINNHUB_KEYS = [key for key in FINNHUB_KEYS if key.strip()]
-GOOGLE_API_KEYS = [key for key in GOOGLE_API_KEYS if key.strip()]
 
 # Validate that we have at least one key for each service
 if not FINNHUB_KEYS:
     print("⚠️ Warning: No Finnhub API keys found in secrets")
 if not RAPIDAPI_KEYS:
     print("⚠️ Warning: No RapidAPI keys found in secrets")
-if not GOOGLE_API_KEYS:
-    print("⚠️ Warning: No Google API keys found in secrets")
-
-# Chọn ngẫu nhiên một khóa API để bắt đầu (if available)
-GOOGLE_API_KEY = random.choice(GOOGLE_API_KEYS) if GOOGLE_API_KEYS else None
+if not HF_TOKEN:
+    print("⚠️ Warning: No Hugging Face token found in secrets")
 
 print("=" * 50)
 print("🚀 FinRobot Forecaster Starting Up...")
@@ -76,20 +73,48 @@ if RAPIDAPI_KEYS:
     print(f"📈 RapidAPI Alpha Vantage: {RAPIDAPI_HOST} ({len(RAPIDAPI_KEYS)} keys loaded)")
 else:
     print("📈 RapidAPI Alpha Vantage: Not configured")
-if GOOGLE_API_KEYS:
-    print(f"🤖 Google Gemini API: {len(GOOGLE_API_KEYS)} keys loaded")
+if HF_TOKEN:
+    print(f"🤖 FinGPT Model: {FINGPT_MODEL_NAME} loaded")
 else:
-    print("🤖 Google Gemini API: Not configured")
+    print("🤖 FinGPT Model: Not configured")
 print("✅ Application started successfully!")
 print("=" * 50)
 
-# Cấu hình Google Generative AI (if keys available)
-if GOOGLE_API_KEYS:
-    # Configure with first key for initial setup
-    genai.configure(api_key=GOOGLE_API_KEYS[0])
-    print(f"✅ Google AI configured with {len(GOOGLE_API_KEYS)} keys")
+# Initialize FinGPT model (if token available)
+if HF_TOKEN:
+    try:
+        # Configure BitsAndBytesConfig for memory efficiency
+        bnb_config = BitsAndBytesConfig(
+            load_in_4bit=True,
+            bnb_4bit_use_double_quant=True,
+            bnb_4bit_quant_type="nf4",
+            bnb_4bit_compute_dtype=torch.bfloat16
+        )
+        
+        # Load tokenizer and model
+        tokenizer = AutoTokenizer.from_pretrained(
+            FINGPT_MODEL_NAME,
+            token=HF_TOKEN,
+            trust_remote_code=True
+        )
+        
+        model = AutoModelForCausalLM.from_pretrained(
+            FINGPT_MODEL_NAME,
+            token=HF_TOKEN,
+            quantization_config=bnb_config,
+            device_map="auto",
+            trust_remote_code=True
+        )
+        
+        print(f"✅ FinGPT model {FINGPT_MODEL_NAME} loaded successfully")
+    except Exception as e:
+        print(f"❌ Error loading FinGPT model: {e}")
+        model = None
+        tokenizer = None
 else:
-    print("⚠️ Google AI not configured - will use mock responses")
+    print("⚠️ FinGPT not configured - will use mock responses")
+    model = None
+    tokenizer = None
 
 # Cấu hình Finnhub client (if keys available)
 if FINNHUB_KEYS:
@@ -413,62 +438,53 @@ def make_prompt(symbol: str, df: pd.DataFrame, curday: str, use_basics=False) ->
 # ---------- LLM CALL -------------------------------------------------------
 
 def chat_completion(prompt: str,
-                    model: str = GEMINI_MODEL,
-                    temperature: float = 0.2,
+                    model_name: str = FINGPT_MODEL_NAME,
+                    temperature: float = 0.7,
                     stream: bool = False,
                     symbol: str = "STOCK") -> str:
-    # Check if Google API keys are configured
-    if not GOOGLE_API_KEYS:
-        print(f"⚠️ Google API not configured, using mock response for {symbol}")
+    # Check if FinGPT model is configured
+    if model is None or tokenizer is None:
+        print(f"⚠️ FinGPT model not configured, using mock response for {symbol}")
         return create_mock_ai_response(symbol)
     
-    # Thử với tất cả các Google API keys
-    for api_key in GOOGLE_API_KEYS:
-        try:
-            # Cấu hình lại với key hiện tại
-            genai.configure(api_key=api_key)
-            
-            # Tạo instance của model
-            model_instance = genai.GenerativeModel(
-                model_name=model,
-                generation_config={
-                    "max_output_tokens": 6144,
-                    "temperature": temperature,
-                    "top_p": 0.9,
-                    "top_k": 40,
-                }
+    try:
+        # Kết hợp system prompt và user prompt
+        full_prompt = f"{SYSTEM_PROMPT}\n\n{prompt}"
+        
+        # Tokenize input
+        inputs = tokenizer.encode(full_prompt, return_tensors="pt", max_length=MAX_LENGTH, truncation=True)
+        
+        # Generate response
+        with torch.no_grad():
+            outputs = model.generate(
+                inputs,
+                max_length=inputs.shape[1] + 512,  # Generate up to 512 new tokens
+                temperature=temperature,
+                do_sample=True,
+                top_p=0.9,
+                top_k=50,
+                pad_token_id=tokenizer.eos_token_id,
+                eos_token_id=tokenizer.eos_token_id,
+                no_repeat_ngram_size=2,
+                early_stopping=True
             )
-
-            # Kết hợp system prompt và user prompt
-            full_prompt = f"{SYSTEM_PROMPT}\n\n{prompt}"
-
-            if stream:
-                response = model_instance.generate_content(full_prompt, stream=True)
-                collected = []
-                for chunk in response:
-                    if chunk.text:
-                        print(chunk.text, end="", flush=True)
-                        collected.append(chunk.text)
-                print()
-                return "".join(collected)
-            else:
-                response = model_instance.generate_content(full_prompt)
-                return response.text
-                
-        except Exception as e:
-            print(f"Error with Google API key {api_key[:10]}...: {e}")
-            if "quota" in str(e).lower() or "limit" in str(e).lower():
-                print(f"Quota/limit hit with key {api_key[:10]}..., trying next key")
-                continue
-            # Nếu không phải lỗi quota, thử key tiếp theo
-            continue
-    
-    # Fallback: Tạo mock AI response khi tất cả Google API keys đều fail
-    print("⚠️ All Google API keys failed, using mock AI response for demonstration...")
-    return create_mock_ai_response(symbol)
+        
+        # Decode response
+        response = tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True)
+        
+        # Clean up response
+        response = response.strip()
+        if not response:
+            return create_mock_ai_response(symbol)
+            
+        return response
+        
+    except Exception as e:
+        print(f"Error with FinGPT model: {e}")
+        return create_mock_ai_response(symbol)
 
 def create_mock_ai_response(symbol: str) -> str:
-    """Tạo mock AI response khi Google API không hoạt động"""
+    """Tạo mock AI response khi FinGPT model không hoạt động"""
     return f"""
 [Positive Developments]
 • Strong market position and brand recognition for {symbol}
@@ -582,9 +598,9 @@ def create_interface():
         gr.Markdown("""
         # 🤖 FinRobot Forecaster
         
-        **AI-powered stock market analysis and prediction using advanced language models**
+        **AI-powered stock market analysis and prediction using FinGPT financial language model**
         
-        This application analyzes stock market data, company news, and financial metrics to provide comprehensive market insights and predictions.
+        This application analyzes stock market data, company news, and financial metrics using the specialized FinGPT model to provide comprehensive market insights and predictions.
         
         ⚠️ **Note**: Free API keys have daily rate limits. If you encounter errors, the app will use mock data for demonstration purposes.
         """)

requirements.txt

@@ -1,10 +1,14 @@
 gradio==4.44.0
 pandas>=1.5.0
 finnhub-python>=2.4.0
-google-generativeai>=0.3.0
 requests>=2.28.0
 urllib3>=1.26.0
 numpy>=1.21.0
 matplotlib>=3.5.0
 plotly>=5.0.0
 yfinance>=0.2.0
+transformers>=4.30.0
+torch>=2.0.0
+accelerate>=0.20.0
+peft>=0.4.0
+bitsandbytes>=0.39.0

test_fingpt_integration.py

@@ -0,0 +1,77 @@
+#!/usr/bin/env python3
+"""
+Test script to verify FinGPT integration
+"""
+
+import os
+import sys
+
+def test_imports():
+    """Test if all required imports work"""
+    try:
+        import torch
+        from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+        from peft import PeftModel
+        print("✅ All FinGPT imports successful")
+        return True
+    except ImportError as e:
+        print(f"❌ Import error: {e}")
+        return False
+
+def test_model_config():
+    """Test model configuration"""
+    try:
+        # Test if we can access the model configuration
+        FINGPT_MODEL_NAME = "Starfish55/fingpt-complete"
+        print(f"✅ Model configuration: {FINGPT_MODEL_NAME}")
+        return True
+    except Exception as e:
+        print(f"❌ Model configuration error: {e}")
+        return False
+
+def test_app_import():
+    """Test if the main app can be imported"""
+    try:
+        # Add current directory to path
+        sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+        
+        # Import the main app
+        import app
+        print("✅ App import successful")
+        return True
+    except Exception as e:
+        print(f"❌ App import error: {e}")
+        return False
+
+def main():
+    """Run all tests"""
+    print("🧪 Testing FinGPT Integration...")
+    print("=" * 50)
+    
+    tests = [
+        test_imports,
+        test_model_config,
+        test_app_import
+    ]
+    
+    passed = 0
+    total = len(tests)
+    
+    for test in tests:
+        if test():
+            passed += 1
+        print()
+    
+    print("=" * 50)
+    print(f"📊 Test Results: {passed}/{total} tests passed")
+    
+    if passed == total:
+        print("🎉 All tests passed! FinGPT integration is ready.")
+    else:
+        print("⚠️ Some tests failed. Please check the errors above.")
+    
+    return passed == total
+
+if __name__ == "__main__":
+    success = main()
+    sys.exit(0 if success else 1)