Anton Troynikov
Anton Troynikov
something like that yeah
I don't think I agree. To use the batching utils a user has to, in addition to the regular code: ```python from chromadb.utils.batch_utils import create_batches ... for (ids, embeddings) in...
In the case of a partially failed batch, the simplest thing to do would be assume the user would run the same operation again; in the case of `.add`, IDs...
The thing we're trying to help / solve for users is that one of the first things they want to do is load a bunch of data into Chroma, and...
Providing a batching utility might be nice for users who do want more control themselves, but my aim with this feature is for the default experience to be much better.
How do you mean 'take over the terminal' ? How is this different to regular log output except that if written to a file it looks weird?
rich looks OK to me
Discussed offline; - will implement this as `batch_upsert`/ `batch_add` for now, punting on what to do with auto-batching for `add`. - we can hoist this to be the default way...
Users may want to change things about the persisted embedding function which don't change the embedding output itself. For example, users might want to update / rotate their API keys....
# Evaluating Snowflake Arctic Embed under llama.cpp I've been investigating whether we ought to make the switch to Snowflake's arctic models as our default embedding model. Currently they have the...