api/QUICK_START.md

🚀 Quick Start Guide - Crypto API Monitor with HuggingFace Integration

✅ Server is Running!

Your application is now live at: http://localhost:7860

📱 Access Points

1. Main Dashboard (Full Features)

URL: http://localhost:7860/index.html

Features:

• Real-time API monitoring
• Provider inventory
• Rate limit tracking
• Connection logs
• Schedule management
• Data freshness monitoring
• Failure analysis
• 🤗 HuggingFace Tab (NEW!)

2. HuggingFace Console (Standalone)

URL: http://localhost:7860/hf_console.html

Features:

• HF Health Status
• Models Registry Browser
• Datasets Registry Browser
• Local Search (snapshot)
• Sentiment Analysis (local pipeline)

3. API Documentation

URL: http://localhost:7860/docs

Interactive API documentation with all endpoints

🤗 HuggingFace Features

Available Endpoints:

1. Health Check

   GET /api/hf/health
   

   Returns: Registry health, last refresh time, model/dataset counts

2. Force Refresh Registry

   POST /api/hf/refresh
   

   Manually trigger registry update from HuggingFace Hub

3. Get Models Registry

   GET /api/hf/registry?kind=models
   

   Returns: List of all cached crypto-related models

4. Get Datasets Registry

   GET /api/hf/registry?kind=datasets
   

   Returns: List of all cached crypto-related datasets

5. Search Registry

   GET /api/hf/search?q=crypto&kind=models
   

   Search local snapshot for models or datasets

6. Run Sentiment Analysis

   POST /api/hf/run-sentiment
   Body: {"texts": ["BTC strong", "ETH weak"]}
   

   Analyze crypto sentiment using local transformers

🎯 How to Use

Option 1: Main Dashboard

1. Open http://localhost:7860/index.html in your browser
2. Click on the "🤗 HuggingFace" tab at the top
3. Explore:
   • Health status
   • Models and datasets registries
   • Search functionality
   • Sentiment analysis

Option 2: Standalone HF Console

1. Open http://localhost:7860/hf_console.html
2. All HF features in a clean, focused interface
3. Perfect for testing and development

🧪 Test the Integration

Test 1: Check Health

Invoke-WebRequest -Uri "http://localhost:7860/api/hf/health" -UseBasicParsing | Select-Object -ExpandProperty Content

Test 2: Refresh Registry

Invoke-WebRequest -Uri "http://localhost:7860/api/hf/refresh" -Method POST -UseBasicParsing | Select-Object -ExpandProperty Content

Test 3: Get Models

Invoke-WebRequest -Uri "http://localhost:7860/api/hf/registry?kind=models" -UseBasicParsing | Select-Object -ExpandProperty Content

Test 4: Run Sentiment Analysis

$body = @{texts = @("BTC strong breakout", "ETH looks weak")} | ConvertTo-Json
Invoke-WebRequest -Uri "http://localhost:7860/api/hf/run-sentiment" -Method POST -Body $body -ContentType "application/json" -UseBasicParsing | Select-Object -ExpandProperty Content

📊 What's Included

Seed Models (Always Available):

• ElKulako/cryptobert
• kk08/CryptoBERT

Seed Datasets (Always Available):

• linxy/CryptoCoin
• WinkingFace/CryptoLM-Bitcoin-BTC-USDT
• WinkingFace/CryptoLM-Ethereum-ETH-USDT
• WinkingFace/CryptoLM-Solana-SOL-USDT
• WinkingFace/CryptoLM-Ripple-XRP-USDT

Auto-Discovery:

• Searches HuggingFace Hub for crypto-related models
• Searches for sentiment-analysis models
• Auto-refreshes every 6 hours (configurable)

⚙️ Configuration

Edit .env file to customize:

# HuggingFace Token (optional, for higher rate limits)
HUGGINGFACE_TOKEN=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

# Enable/disable local sentiment analysis
ENABLE_SENTIMENT=true

# Model selection
SENTIMENT_SOCIAL_MODEL=ElKulako/cryptobert
SENTIMENT_NEWS_MODEL=kk08/CryptoBERT

# Refresh interval (seconds)
HF_REGISTRY_REFRESH_SEC=21600

# HTTP timeout (seconds)
HF_HTTP_TIMEOUT=8.0

🛑 Stop the Server

Press CTRL+C in the terminal where the server is running

Or use the process manager to stop process ID 6

🔄 Restart the Server

python simple_server.py

📝 Notes

• First Load: The first sentiment analysis may take 30-60 seconds as models download
• Registry: Auto-refreshes every 6 hours, or manually via the UI
• Free Resources: All endpoints use free HuggingFace APIs
• No API Key Required: Works without authentication (with rate limits)
• Local Inference: Sentiment analysis runs locally using transformers

🎉 You're All Set!

The application is running and ready to use. Open your browser and explore!

Main Dashboard: http://localhost:7860/index.html HF Console: http://localhost:7860/hf_console.html