Spaces:
Configuration error
Configuration error
| """Hugging Face client - Inference (HTTP router + Inference Endpoints) and hub. | |
| Uses huggingface_hub when present (pip install huggingface_hub), else raw HTTP. | |
| """ | |
| import os | |
| import requests | |
| ROUTER = os.getenv("HF_INFERENCE_API", "https://router.huggingface.co") | |
| TOKEN = os.getenv("HF_TOKEN") | |
| def _h(): | |
| if not TOKEN: | |
| raise RuntimeError("HF_TOKEN not set") | |
| return {"Authorization": f"Bearer {TOKEN}", "Content-Type": "application/json"} | |
| def chat_completion(model, messages, **kwargs): | |
| return requests.post(f"{ROUTER}/v1/chat/completions", headers=_h(), | |
| json={"model": model, "messages": messages, **kwargs}, timeout=120).json() | |
| def text_generation(model, prompt, **kwargs): | |
| return requests.post(f"{ROUTER}/models/{model}", headers=_h(), | |
| json={"inputs": prompt, **kwargs}, timeout=120).json() | |
| def list_models(): | |
| return requests.get(f"{ROUTER}/v1/models", headers=_h(), timeout=30).json() | |
| def infer_endpoint(endpoint_id: str, payload: dict): | |
| """Query a dedicated Inference Endpoint (set HF_INFERENCE_ENDPOINT base).""" | |
| base = os.getenv("HF_INFERENCE_ENDPOINT", ROUTER) | |
| return requests.post(f"{base}/{endpoint_id}", headers=_h(), json=payload, timeout=120).json() | |
| def whoami(): | |
| """Hub identity via huggingface_hub (optional dep).""" | |
| try: | |
| from huggingface_hub import whoami as _whoami | |
| return _whoami() | |
| except ImportError: | |
| return {"note": "install huggingface_hub for hub identity/upload APIs"} | |