asagi4
|
6d50f33811
|
I don't think -1 is a useful value for this
|
2025-06-23 20:54:34 +03:00 |
|
asagi4
|
4335902543
|
Add seed to CADS
|
2025-06-23 20:54:30 +03:00 |
|
asagi4
|
a27be21f81
|
Test noising the prediction instead
|
2025-06-23 20:40:27 +03:00 |
|
asagi4
|
9d114ffdaa
|
Add support for different kinds of noise and relax noise_scale limits
|
2024-04-08 18:51:22 +03:00 |
|
blepping
|
b143f4eb53
|
Change step state handling to avoid having to force a workflow rerun
|
2024-03-24 06:24:56 -06:00 |
|
blepping
|
b77cee21a6
|
Change variable name to avoid confusion with apply_to
|
2024-03-24 04:06:24 -06:00 |
|
blepping
|
9b2a19eef9
|
Fall back to using c_crossattn if y key does not exist
|
2024-03-24 04:01:18 -06:00 |
|
asagi4
|
42a84c7874
|
Cleanup
|
2024-03-10 02:17:44 +02:00 |
|
asagi4
|
b48bc94c94
|
Default apply_to to uncond since that seems to make the most sense
|
2024-03-10 02:14:26 +02:00 |
|
asagi4
|
c8567a7783
|
Apply noise to y instead of c_crossattn
|
2024-03-10 01:53:17 +02:00 |
|
asagi4
|
8fcef78cba
|
Small cleanup. a NaN should result in the node always re-executing
|
2024-01-06 21:49:16 +02:00 |
|
asagi4
|
eec1490710
|
Allow skipping CADS for either cond or uncond
|
2023-12-06 19:35:29 +02:00 |
|
asagi4
|
59d67f304b
|
Fixes
|
2023-12-06 14:07:48 +02:00 |
|
asagi4
|
e4bcfba98e
|
Reformat
|
2023-12-06 13:24:46 +02:00 |
|
asagi4
|
9752ff3f25
|
Initial commit
|
2023-12-06 13:24:30 +02:00 |
|