Commit
·
920e420
1
Parent(s):
e5df751
Update README.md
Browse files
README.md
CHANGED
|
@@ -44,13 +44,12 @@ license: creativeml-openrail-m
|
|
| 44 |
The Protogen Nova is a checkpoint model that merges all the previous models into one
|
| 45 |
|
| 46 |
This merger includes
|
| 47 |
-
* Protogen v2.2
|
| 48 |
-
* Protogen x3.4
|
| 49 |
-
* ProtoGen x5.3
|
| 50 |
-
* ProtoGen x5.8
|
| 51 |
-
* ProtoGen x5.9
|
| 52 |
-
* ProtoGen x7.4
|
| 53 |
-
* ProtoGen x7.9
|
| 54 |
|
| 55 |
As part the of the checkpoint merging, Granular Adaptive Learning is a technique where traininig data is lessen selectively from 30% to 0.05%, and as the training is eventually saturated, the process reduces loss and introduces elements from various checkpoints
|
| 56 |
|
|
|
|
| 44 |
The Protogen Nova is a checkpoint model that merges all the previous models into one
|
| 45 |
|
| 46 |
This merger includes
|
| 47 |
+
* Protogen v2.2 (Anime)
|
| 48 |
+
* Protogen x3.4 (Photorealism)
|
| 49 |
+
* ProtoGen x5.3 (Photorealism)
|
| 50 |
+
* ProtoGen x5.8 Rebuilt (Scifi+Anime)
|
| 51 |
+
* ProtoGen x5.9 (Dragon)
|
| 52 |
+
* ProtoGen x7.4 (Eclipse)
|
|
|
|
| 53 |
|
| 54 |
As part the of the checkpoint merging, Granular Adaptive Learning is a technique where traininig data is lessen selectively from 30% to 0.05%, and as the training is eventually saturated, the process reduces loss and introduces elements from various checkpoints
|
| 55 |
|