JamieYuu/crypto-news-custom-endpoint

Crypto News Custom Inference Endpoint

This repo is endpoint-ready for custom multi-input inference:

• text (news)
• btc_price_now
• fng_value
• fng_classification

Output fields:

• pred_class
• sentiment
• score
• prob_up
• confidence

Deploy on Hugging Face Inference Endpoints

1. Go to Inference Endpoints and create a new endpoint from this model repo.
2. Task/engine: use custom handler mode (repository contains handler.py).
3. Ensure environment installs requirements.txt.
4. Deploy.

API Payload Example

{
  "inputs": {
    "text": "Bitcoin ETF inflows rose this week as institutional demand accelerated.",
    "btc_price_now": 67000,
    "fng_value": 62,
    "fng_classification": "Greed"
  }
}

Python Snippet

import requests

ENDPOINT_URL = "https://YOUR_ENDPOINT_URL"
TOKEN = "hf_xxx"

payload = {
    "inputs": {
        "text": "Bitcoin ETF inflows rose this week as institutional demand accelerated.",
        "btc_price_now": 67000,
        "fng_value": 62,
        "fng_classification": "Greed"
    }
}

resp = requests.post(
    ENDPOINT_URL,
    headers={"Authorization": f"Bearer {TOKEN}"},
    json=payload,
    timeout=60,
)
resp.raise_for_status()
print(resp.json())