Add metadata as a field in the MemoryRecord class in Python #1271
Comments
cschadewitz
changed the title
evchaki
added
python
|
Good catch! We will add. This is already in C# so Python needs to catch up |
|
I see that you assigned to @awharrison-28 but if youd like I would be happy to work on this along side the pinecone memory store implementation that I have taken over and the postgres memory store implementation that I have all but finished. Alternatively, I would also be happy to just collaborate. Either way, I'll prep branches for changes against those memory stores to go in after this feature. |
|
@cschadewitz thanks for the offer. This is a really quick one to put out, so I'll take care of it asap. |
|
@cschadewitz PR is live, please take a look to make sure it covers what you need for Pinecone and Postgres :) |
|
@awharrison-28 I'll take a look this evening and let you know if I have any suggestions. Thanks for getting this in place so quickly! |
|
@awharrison-28, sorry it took so long for me to get to this but it looks good to me! Thanks again! |
…Os in ChromaMemoryStore (#1323) ### Motivation and Context Closes #1271 and addresses several TODOs in ChromaMemoryStore ### Description - added field `additional_metadata` to `MemoryRecord` and all downstream methods/classes (see `SemanticTextMemoryBase`, `SemanticTextMemory`, `MemoryQueryResult`, and `ChromaMemoryStore`) - addressed ChromaMemoryStore TODO 'what is key used for?' For chromadb, key and id are effectively the same value, but the two fields are differentiated for memory stores like Qdrant which require that key be a uuid. I updated the logic so that this is more apparent. Note that this results in no change in end-user functionality. - Updated ChromaMemoryStore TODO 'get_async say embedding is optional but Record constructor requires it.` MemoryRecord constructor no longer requires an embedding. Logic is updated and a test added to show that embedding retrieval is functionally optional.
…Os in ChromaMemoryStore (microsoft#1323) ### Motivation and Context Closes microsoft#1271 and addresses several TODOs in ChromaMemoryStore ### Description - added field `additional_metadata` to `MemoryRecord` and all downstream methods/classes (see `SemanticTextMemoryBase`, `SemanticTextMemory`, `MemoryQueryResult`, and `ChromaMemoryStore`) - addressed ChromaMemoryStore TODO 'what is key used for?' For chromadb, key and id are effectively the same value, but the two fields are differentiated for memory stores like Qdrant which require that key be a uuid. I updated the logic so that this is more apparent. Note that this results in no change in end-user functionality. - Updated ChromaMemoryStore TODO 'get_async say embedding is optional but Record constructor requires it.` MemoryRecord constructor no longer requires an embedding. Logic is updated and a test added to show that embedding retrieval is functionally optional.
Add metadata as a field in the python memory record class.
This would improve parity between the implementations.
The field would be a dict[str, str | int | bool | list[str]] this would allow for a flexible usage of the field and would mesh well with Pinecone's db record metadata implementation and could be stored in other dbs as Json or JSONb
Furthermore, the text, description and other such fields on the python memory record class could be moved to this dict as well similar to the .net Memory Record class
The text was updated successfully, but these errors were encountered: