VatsaDev's repositories
NanoPhi-alpha
GPT-2 small trained on phi-like data
layersVdimension
wider layers vs more layers
simple-evo
Chickens hunted by Foxes, ML
TransformerMath
Can transformers learn math, like patterns?
NCPT-Lilith
A retrain of the old nanogpt, but with the lilith optimizer
vatsadev.github.io
website
curriculum-experiment
testing out a curriculum learning method
fact-interp
Trying out interp and transformers abilities with random facts missing
Language:HTMLMIT000
Special-topics-logs
Keep track of the stuff I've done for special topics