8000 History for tools - drdaxxy/dalle-mini · GitHub

Repository navigation

Commits

History for

Commits on Jun 17, 2022
- feat(colab): allow custom project name (#264)
  borisdayma
  authored
  791b1ea
- feat(colab): add W&B Tables (#259)
  ayulockin
  authored
  6f585d4
Commits on Jun 9, 2022
- feat(docker): improve usage of Docker and infer notebook (#218)
  mallorbc
  authored
  7feb906
Commits on Jun 7, 2022
- feat(demo): allow multiple prompts
  borisdayma
  committed
  e58bdf2
- docs: Improve readability in `.py` files (#209)
  ryanrussell
  authored
  66d8a92
Commits on May 31, 2022
- feat: support multi validation datasets (#192)
  borisdayma
  authored
3109050

Commits on May 28, 2022

feat(colab): increase conditioning scale
borisdayma
authored
d12a529

Commits on May 23, 2022

feat(train): do eval in fp32 (#190)
borisdayma
authored
403182b

Commits on May 10, 2022

fix(colab): use correct param name for CLIP
borisdayma
committed
badd15c

Commits on May 9, 2022

feat(colab): use full precision for clip & vqgan
borisdayma
committed
8845d77
feat(colab): require less memory
borisdayma
committed
5f16fb0

Commits on May 5, 2022

fix(train): overwrite dropout only when specified
borisdayma
committed
89bc9d4
feat: allow eval only
borisdayma
committed
65bb95f

Commits on May 3, 2022

feat(colab): update mini checkpoint
borisdayma
committed
a4d2af8
feat(train): allow editing dropout during training
borisdayma
committed
80d791a
feat(colab): require less resources
borisdayma
committed
b6f5026

Commits on May 2, 2022

feat(colab): change defaults
borisdayma
committed
3de3769
feat(colab): handle dalle-mega
borisdayma
committed
024f8f5

Commits on Apr 29, 2022

feat: support LR offset (#174)
borisdayma
authored
c3e93df
feat: update shampoo
borisdayma
committed
79a3849
feat(train): allow nesterov momentum
borisdayma
committed
d08bf8d

Commits on Apr 28, 2022

feat: allow weight decay
borisdayma
committed
acc1a4a

Commits on Apr 20, 2022

style: reformat
borisdayma
committed
dca3ada

Commits on Apr 19, 2022

feat(train): use new HF _do_init api
borisdayma
committed
6b84155
fix: model compatible with do_init
borisdayma
committed
f3a8cbb

Commits on Apr 16, 2022

fix: apply learning rate offset only when requested
borisdayma
committed
c6263f3
feat(train): arg to offset lr for resumed runs
borisdayma
committed
89b4c45

Commits on Apr 15, 2022

Commits on Apr 11, 2022

feat: vmap optimizer (#166)
borisdayma
authored
b993d27
feat(demo): use fixed commit
borisdayma
committed
2f1e5d9

Commits on Apr 9, 2022

feat: scan layers + gradient checkpointing (#161)
borisdayma
authored
07a6f9a

Commits on Apr 8, 2022

feat: better multi-node support (#158)
borisdayma
authored
728a3c3

0