Kevin Black
|
1958463f02
|
Reformat
|
2023-11-16 22:37:08 +00:00 |
|
Kevin Black
|
5955244f37
|
Fix gradient sync for lora
|
2023-08-22 23:34:40 +00:00 |
|
Desai Xie
|
3130ddfaff
|
Only log rewards from process 0
|
2023-08-21 15:10:45 -07:00 |
|
Kevin Black
|
ec499edf84
|
Fix aesthetic score (again), add llava reward
|
2023-07-04 00:23:33 -07:00 |
|
Kevin Black
|
c0bc708549
|
Commenting pass
|
2023-06-29 00:51:38 -07:00 |
|
Kevin Black
|
8779f62a1c
|
Adding checkpointing and resuming
|
2023-06-28 17:58:25 -07:00 |
|
Kevin Black
|
ad28862b48
|
Add reward to image caption
|
2023-06-28 10:42:47 -07:00 |
|
Kevin Black
|
28d2d8c40e
|
Minor changes; add train_timestep_fraction
|
2023-06-27 22:17:56 -07:00 |
|
Kevin Black
|
bae3f43f5f
|
Add aesthetic scorer reward function
|
2023-06-27 10:40:36 -07:00 |
|
Kevin Black
|
8cab96dea4
|
Minor changes, add assets
|
2023-06-27 10:20:03 -07:00 |
|
Kevin Black
|
4c5322ca85
|
Device specific seed
|
2023-06-26 22:35:24 -07:00 |
|
Kevin Black
|
5c16a90ceb
|
Move config out of module
|
2023-06-25 21:02:27 -07:00 |
|
Kevin Black
|
269615a35e
|
Working non-lora training; other changes
|
2023-06-25 11:28:42 -07:00 |
|
Kevin Black
|
c680890d5c
|
Working on DGX
|
2023-06-24 00:07:55 -07:00 |
|
Kevin Black
|
92fc030123
|
Continue implementation
|
2023-06-23 21:08:32 -07:00 |
|
Kevin Black
|
2fda3d4e78
|
Initial commit
|
2023-06-23 19:25:54 -07:00 |
|