Why Does the HuggingFace Embeddings Node on n8n Cloud Return an 'Invalid URL' Error?

It looks like the error is related to the configuration of your custom inference endpoint or the way the URL is provided. Here are a few suggestions to troubleshoot the issue:

1. Validate the Custom Inference Endpoint

• Format: Ensure that the URL is complete with the protocol (e.g., https://).
• If Using Hugging Face Hosted Models:
  If you intend to use one of Hugging Face’s hosted models (like BAAI/bge-small-en-v1.5), try leaving the Custom Inference Endpoint field empty. In this case, the node will use the model selected in the Model parameter and handle the connection automatically.

2. Check Your Credentials

• Verify that your Hugging Face credentials are correctly set up in n8n (i.e., using a valid API key).
• The credentials documentation shows that you need to obtain an API token (starting with hf_) from your Hugging Face account and use that in the node configuration.

3. Confirm the Endpoint Is Accessible

• If you are indeed using a custom endpoint (for example, if you are self-hosting the model), double-check that the URL is reachable and that it points to a valid inference endpoint.
• Confirm that there are no extra spaces or typos in the URL, as these can trigger an Invalid URL error.

4. Review Documentation

• The Embeddings HuggingFace Inference node documentation provides guidance on both the Model selection and the Custom Inference Endpoint option.
• For further details on using Hugging Face endpoints, refer to Hugging Face’s guide to inference endpoints.

By reviewing these configurations and ensuring the URL is correctly specified, you should be able to resolve the Invalid URL error in your workflow.