Closed dlg99 closed 1 year ago
Regarding the fact that we use HG only for embeddings and that the HG configuration should be at the top level like "openai", I think that we could support the "Inference API" at some point in the future. https://huggingface.co/docs/api-inference/detailed_parameters
Having it at the top level also makes the configuration easier for the users
EmbeddingsService implementations for Huggingface
The model requested there should be trained for "sentence similarity" task