Having trouble loading EmbeddingGemma

[rodydavis]

Here is the SQL Code:

create virtual table chunks using vec0(
  id integer primary key,
  embedding float[768]
);

INSERT INTO temp.lembed_models(name, model)
  select 'embeddinggemma', lembed_model_from_file('./models/embeddinggemma-300m-qat-Q4_0.gguf');

select lembed(
  'embeddinggemma',
  'The United States Postal Service is an independent agency...'
);

Getting the error here:

sqlite> INSERT INTO temp.lembed_models(name, model)
  select 'embeddinggemma', lembed_model_from_file('./models/embeddinggemma-300m-qat-Q4_0.gguf');
Runtime error: SQL logic error

I installed the latest version, but I think llama.cpp needs to be updated

[O-J1]

Yes, its severely behind unfortunately

[rodydavis]

I think we should be able to update Llama.cpp

Going to attempt a PR

[rodydavis]

Fixed!

#19

[rodydavis]

EmbeddingGemma now works!