The post indicates that the model_name is set to None if it is not found, which overrides the default in the base class, resulting in a failure. In the comments, a community member expresses confusion about the issue and the error. Another community member shares that their code was initially set to Settings.embed_model = 'local', but suddenly started failing, and they had to change it to Settings.embed_model = 'local:BAAI/bge-small-en' for it to work, which they consider a breaking change. They mention that this can become messy for large projects in the cloud, as their project is related to llama_index.
my code was initially set to Settings.embed_model = 'local' never had to touch it and all of a sudden the entire project started failing and found out the settings code is updated. now i have to change it to Settings.embed_model = 'local:BAAI/bge-small-en' for it to work, so it was a breaking change. thought of letting you know that like kindoff an FYI. It becomes messy for large projects in the cloud as our project is pretty much llama_index 😎
