rj45's learning exhaust

rj45's learning exhaust

Jan 12, 202613 min read

Introducing the AnnData DuckDB extension

At MiraOmics we look at a lot of H5AD files. We have a database of over a thousand of them, not to mention several of the public data sources. Pretty much every time I have a new file, the first thing I want to do is run a few queries on it to under...

Lab Notes: Comparing GenePT and scGPT
In the spirit of “Learning in Public,” and “learning exhaust,” I’m going to start adding “Lab Notes” blog posts that chronicle little discoveries or failures that I have along the way to a larger goal. This is the first! Background I’m collaborating ...
Feb 20, 20259 min read
Data Products are Different
... and AI is a data product
Dec 28, 202410 min read
One thing I learned: weight streaming probably works well on GPUs
I wonder why it isn't a thing...
Dec 13, 20247 min read
One thing I learned: embeddings are task-specific
We should probably include a prompt or fine tune to a specific type of task
Nov 28, 20245 min read
Is it the model or the data that's low rank?
Thoughts on LoRA Quantization Loss Recovery
Aug 11, 20244 min read
Can we improve quantization by fine tuning?
or post-post-training-quantization-training :)
Jul 12, 20248 min read