Spaces:
Running
Running
Commit History
handle dtype for embeddings
29db327
add gradient checkpointing
95a8ed2
fix layernorm
7774483
add partition helpers
2856356
remove bias and minor fixes
180ed1e
feat: reorganize samples
0588e94
feat: add more samples
046ae75
feat: update wandb inference
dc79278
Merge pull request #91 from borisdayma/feat-inf
335110d
unverified
feat: more samples
2ef2966
Merge pull request #90 from borisdayma/feat-new
fdbe19f
unverified
feat: cleanup
91d8a29
feat: add sample
1d51d0b
fix: typo
c85fbb6
fix: pmap clip32
9a553a4
feat: cleanup
2d169e3
feat: allow latest version only
ff051c9
refactor: loop over runs
bf3640d
feat: create a table
38705a9
feat: add scoring
353365f
feat: remove cache before creating artifacts
5f6b691
feat: add functions
b8bbe68
feat: log to backend
378a628
fix: state.step type
47e006f
Merge branch 'main' of https://github.com/borisdayma/dalle-mini
5faf0fd
feat: get rid of global_step + log more metrics
4a4820f
fix(seq2seq): memory issue
708a42c
Merge pull request #88 from borisdayma/feat-cumul
272552a
unverified
add standalone modeling file
6197b2f
feat(gitignore): ignore compiled files and wandb
0b21fa5
feat: update scripts
5ca30e6
feat: use optax for gradient accumulation
69cf636
Merge pull request #71 from borisdayma/fix-opt_state
77657e6
unverified
Pedro Cuenca
commited on