| from huggingface_hub import InferenceClient, create_inference_endpoint | |
| # Create the inference endpoint | |
| endpoint = create_inference_endpoint( | |
| name="my-custom-endpoint", | |
| repository="path/to/your/model/repository", | |
| framework="custom", | |
| task="text-classification", | |
| accelerator="cpu", # or "gpu" if needed | |
| instance_size="medium", | |
| instance_type="c6i", | |
| region="us-east-1", | |
| custom_image={ | |
| "health_route": "/healthz", | |
| "port": 8080, | |
| "url": "your-docker-image-url:latest" | |
| } | |
| ) | |
| # Wait for the endpoint to be ready | |
| endpoint.wait() | |
| # Create a client to interact with the endpoint | |
| client = InferenceClient(endpoint.url) | |
| # Test the endpoint | |
| result = client.text_classification("This is a test input") | |
| print(result) |