Kevin Wu
|
31cc02c2ce
|
Support for running without a config json
|
2022-07-25 19:10:38 +00:00 |
|
Kevin Wu
|
b435ae587c
|
Update some default values
|
2022-07-25 19:03:21 +00:00 |
|
Kevin Wu
|
6503ab2fc0
|
Properly trim by sequence length
|
2022-07-25 18:19:47 +00:00 |
|
Kevin Wu
|
987bc45968
|
More checkpointing
|
2022-07-20 03:25:47 +00:00 |
|
Kevin Wu
|
0529dcf6d1
|
Implement L1/L2 regularization
|
2022-07-19 00:51:00 +00:00 |
|
Kevin Wu
|
a6909e50cb
|
Increase default early stop patience, add logging
|
2022-07-18 18:00:53 +00:00 |
|
Kevin Wu
|
89893001db
|
Allow for specifying modulo to dataset
|
2022-07-18 02:02:25 +00:00 |
|
Kevin Wu
|
bd40c51a7d
|
Properly implement radian L1 loss
|
2022-07-18 01:33:02 +00:00 |
|
Kevin Wu
|
268afe26a6
|
Misc. fixes
|
2022-07-12 19:29:32 +00:00 |
|
Kevin Wu
|
50f35608ae
|
Implement sampling multiple sequences
|
2022-07-12 19:02:43 +00:00 |
|
Kevin Wu
|
edbc419a33
|
Initial sampling code
|
2022-07-12 17:57:35 +00:00 |
|
Kevin Wu
|
510df0935b
|
Support for gradient clipping
|
2022-07-08 19:31:37 +00:00 |
|
Kevin Wu
|
b4f9033e78
|
Code cleanup
|
2022-07-08 19:08:58 +00:00 |
|
Kevin Wu
|
496c8ea3d1
|
Handle json config in train script
|
2022-07-08 19:02:50 +00:00 |
|
Kevin Wu
|
9a67996816
|
Add early stopping
|
2022-07-08 18:04:15 +00:00 |
|
Kevin Wu
|
b26a61c227
|
Fuller lightning integration
|
2022-07-08 17:49:40 +00:00 |
|
Kevin Wu
|
28b09059a0
|
Incorporate pytorch lightning
|
2022-07-08 17:27:10 +00:00 |
|
Kevin Wu
|
8290b7d802
|
Record all params to training loop
|
2022-07-08 16:14:04 +00:00 |
|
Kevin Wu
|
247e4f6baf
|
Report validation loss
|
2022-07-07 20:50:18 +00:00 |
|
Kevin Wu
|
3377a3e8cc
|
Add data splits to cath training script
|
2022-07-07 18:49:16 +00:00 |
|
Kevin Wu
|
923a9080c0
|
Doc updates
|
2022-07-07 18:11:24 +00:00 |
|
Kevin Wu
|
175ba7adfc
|
More configurability for variance schedules
|
2022-07-07 18:06:11 +00:00 |
|
Kevin Wu
|
19ce311895
|
reproducibility
|
2022-07-07 17:58:39 +00:00 |
|
Kevin Wu
|
71d974db4f
|
Add plot of losses
|
2022-07-07 16:34:23 +00:00 |
|
Kevin Wu
|
93b2be6edf
|
Implement attention mask
|
2022-07-07 02:11:50 +00:00 |
|
Kevin Wu
|
31900261b6
|
Update dataset return values to always be dict
|
2022-07-06 23:11:03 +00:00 |
|
Kevin Wu
|
97dbe9824e
|
Up the batch size!
|
2022-07-06 22:54:45 +00:00 |
|
Kevin Wu
|
e7a4d8c12b
|
Configurable training loop
|
2022-07-06 22:41:46 +00:00 |
|
Kevin Wu
|
b120009d5e
|
Initial training script
|
2022-07-06 22:28:26 +00:00 |
|