Oufattole/meds-torch

Explore different numerical Value embedding strategies

Opened this issue · 2 comments

We currently only support Continuous value embeddings (a one to many FFN). We should try other things, like supporting quantizing.

follow the hydra target schema:

priority tag specifically commenting on comparing ways to turn a coninuous value (code-specific or code-independent) into a vector embedding (e.g., scaling a code-independent embedding vector, scaling a code-specific embedding vector, using an FFN, etc).