Skip to content

Activity

Showing most recent first

Update README.md

lapp0pushed 1 commit to master • 5ca3b5b…eb5f9b9 • 
on Jan 27

Update README.md

lapp0pushed 1 commit to master • 71578d5…5ca3b5b • 
on Jan 27

Update README.md

lapp0pushed 1 commit to master • 81d51b7…71578d5 • 
on Jan 27

misc model.py improvements

lapp0pushed 1 commit to master • e0f8577…81d51b7 • 
on Jan 21

fix compilation of loss calculation in seq classification finetuner b…

lapp0pushed 1 commit to master • d84ff3e…e0f8577 • 
on Jan 21

use position IDs for sliding window mask

lapp0pushed 15 commits to master • 0205bca…d84ff3e • 
on Jan 21

update requirements

lapp0pushed 2 commits to master • cf87959…0205bca • 
on Jan 14

update readme

lapp0pushed 1 commit to master • a64c263…cf87959 • 
on Jan 14

update parameters to be appropriate for long train run

lapp0pushed 1 commit to master • 830990a…a64c263 • 
on Jan 14

update model dim

lapp0pushed 3 commits to master • 29a9816…830990a • 
on Jan 14

Refactor trainer as class, (same loss, same memory usage, 99.9% the p…

lapp0pushed 8 commits to master • 20e0260…29a9816 • 
on Jan 14

update readme

lapp0pushed 1 commit to master • 3640026…20e0260 • 
on Jan 12

no need for ignore_keys as nn.Buffer isn't persistent

lapp0pushed 6 commits to master • 1e04c14…3640026 • 
on Jan 12

update readme

lapp0pushed 2 commits to master • 6fd5157…1e04c14 • 
on Jan 12

update readme

lapp0pushed 1 commit to master • 437f2bf…6fd5157 • 
on Jan 12

add related works to readme

lapp0pushed 1 commit to master • c479792…437f2bf • 
on Jan 12

remove runs from readme

lapp0pushed 6 commits to master • 9c44215…c479792 • 
on Jan 12

refactor unet handling

lapp0pushed 15 commits to master • b84ec6d…9c44215 • 
on Jan 11

misc fixes to get mnli finetuner working

lapp0pushed 3 commits to master • 6cace21…b84ec6d • 
on Jan 8

add lr_head setting

lapp0pushed 7 commits to master • 92eef50…6cace21 • 
on Jan 8

separate model / tokenizer from train() script

lapp0pushed 1 commit to master • a5598ca…92eef50 • 
on Jan 6

shrink mlp per SpeedRunningESM2 and ModernBERT

lapp0pushed 1 commit to master • b8f528b…a5598ca • 
on Jan 6

update readme

lapp0pushed 1 commit to master • 43e9f41…b8f528b • 
on Jan 6

hardcode repo URI for now because I keep forgetting...

lapp0pushed 10 commits to master • 8b74a25…43e9f41 • 
on Jan 6

Merge remote-tracking branch 'origin/master'

lapp0pushed 15 commits to master • 3928f38…8b74a25 • 
on Jan 5

Update README.md

lapp0pushed 1 commit to master • 6c1ae3d…3928f38 • 
on Jan 5

misc fixes

lapp0pushed 2 commits to master • 5a72a03…6c1ae3d • 
on Jan 4

cleanup

lapp0pushed 1 commit to master • 6b8b990…5a72a03 • 
on Jan 4

update mlm lerp params - was incorrect

lapp0pushed 1 commit to master • 6e3e92d…6b8b990 • 
on Jan 4

update mlm lerp params

lapp0pushed 1 commit to master • d6c2f9c…6e3e92d • 
on Jan 4