LoreonLabsPlatform

Overview

Intelligence

Markets
Builders
Research
Ecosystems
Launchpads

Search

Other

parameter-golf

Train the smallest LM you can that fits in 16MB. Best model wins!

OtherEmerging

Stars

—

Forks

—

Contributors

8

Last push

3mo ago

Recent commits

Latest commits.

Add record: Sliding Window Eval (stride=64), val_bpb=1.1925 (#50)
d84a3e8Matthew Li3mo ago
Update README.md
194bb87Will DePue3mo ago
SOTA attempt (val_bpb=1.2064) (#49)
e89fcf8spokane-way3mo ago
clarify torch version
6e3e90dAlex3mo ago
Update README.md (#105)
b5ea566Will DePue3mo ago

fp16 tied embedding + lr/warmdown tuning — val_bpb 1.2197 (#42)

a5eb9edRenier Velazco3mo ago

Merge pull request #100 from sandsevenone/mlx_eager_eval

2081ba1Will DePue3mo ago

Add MLX_EAGER_EVAL flag to further reduce memory pressure by force-evaluating the graph after each sub-batch step

6a08c9dsandrone3mo ago

Top contributors

Builders behind this project.