khanon commited on
Commit
0c5fd56
1 Parent(s): acdbfd2

updates README for several characters

Browse files
arona/README.md CHANGED
@@ -34,7 +34,6 @@ Not all tags may be necessary.
34
  ## Training
35
  *Exact parameters are provided in the accompanying JSON files.*
36
  - Trained on a set of 250 images.
37
- - Datset included a mixture of SFW/NSFW
38
  - 5 repeats
39
  - 3 batch size, 4 epochs
40
  - `(250 * 5) / 3 * 4` = 1667 steps
 
34
  ## Training
35
  *Exact parameters are provided in the accompanying JSON files.*
36
  - Trained on a set of 250 images.
 
37
  - 5 repeats
38
  - 3 batch size, 4 epochs
39
  - `(250 * 5) / 3 * 4` = 1667 steps
mari/README.md CHANGED
@@ -34,7 +34,6 @@ For her swimsuit: `black one-piece swimsuit, frilled swimsuit, frills, paw print
34
  ## Training
35
  *Exact parameters are provided in the accompanying JSON files.*
36
  - Trained on a set of 253 images
37
- - Datset included a mixture of SFW/NSFW.
38
  - 3 repeats
39
  - 3 batch size, 7 epochs
40
  - `(253*3) / 3 * 7` = 1771 steps
 
34
  ## Training
35
  *Exact parameters are provided in the accompanying JSON files.*
36
  - Trained on a set of 253 images
 
37
  - 3 repeats
38
  - 3 batch size, 7 epochs
39
  - `(253*3) / 3 * 7` = 1771 steps
mutsuki/README.md CHANGED
@@ -3,7 +3,6 @@ This is my first time doing any sort of Stable Diffusion training so I went thro
3
  # Training
4
  *All parameters are provided in the accompanying JSON files.*
5
  - Trained on 138 curated images, repeated 8 times (1104 total images / 3 batchsize = 368 iterations)
6
- - Dataset included a mixture of SFW and NSFW
7
  - I pruned most images with white backgrounds because I felt they might have been negatively impacting my results early on, but in hindsight I think that was bad training parameters instead.
8
  - Dataset was tagged with WD1.4 interrogator. Shuffling was disabled.
9
  - `mutsuki, blue archive` were added to the start of each caption.
 
3
  # Training
4
  *All parameters are provided in the accompanying JSON files.*
5
  - Trained on 138 curated images, repeated 8 times (1104 total images / 3 batchsize = 368 iterations)
 
6
  - I pruned most images with white backgrounds because I felt they might have been negatively impacting my results early on, but in hindsight I think that was bad training parameters instead.
7
  - Dataset was tagged with WD1.4 interrogator. Shuffling was disabled.
8
  - `mutsuki, blue archive` were added to the start of each caption.
reisa/README.md CHANGED
@@ -29,7 +29,6 @@ For her normal outfit: `school uniform, grey shirt, blue jacket, pleated skirt,
29
  ## Training
30
  *Exact parameters are provided in the accompanying JSON files.*
31
  - Trained on a set of 89 images
32
- - Datset included a mixture of SFW/NSFW
33
  - 13 repeats
34
  - 3 batch size, 4 epochs
35
  - `(89*13) / 3 * 4` = 1543 steps
 
29
  ## Training
30
  *Exact parameters are provided in the accompanying JSON files.*
31
  - Trained on a set of 89 images
 
32
  - 13 repeats
33
  - 3 batch size, 4 epochs
34
  - `(89*13) / 3 * 4` = 1543 steps
shizuko/README.md CHANGED
@@ -37,7 +37,6 @@ Her shotgun sling tends to show up on her normal outfit and is a little hard to
37
  ## Training
38
  *Exact parameters are provided in the accompanying JSON files.*
39
  - Trained on a set of 125 images; 88 swimsuit, 37 normal.
40
- - Datset included a mixture of SFW/NSFW
41
  - 11 repeats for normal outfit
42
  - 8 repeats for swimsuit outfit
43
  - 3 batch size, 4 epochs
 
37
  ## Training
38
  *Exact parameters are provided in the accompanying JSON files.*
39
  - Trained on a set of 125 images; 88 swimsuit, 37 normal.
 
40
  - 11 repeats for normal outfit
41
  - 8 repeats for swimsuit outfit
42
  - 3 batch size, 4 epochs
sora/README.md CHANGED
@@ -32,7 +32,6 @@ Weight 1 is suggested.
32
  ## Training
33
  *Exact parameters are provided in the accompanying JSON files.*
34
  - Trained on a set of 128 images
35
- - Datset included a mixture of SFW/NSFW. Sora has more NSFW than most students.
36
  - 9 repeats
37
  - 3 batch size, 4 epochs
38
  - `(128*9) / 3 * 4` = 1536 steps
 
32
  ## Training
33
  *Exact parameters are provided in the accompanying JSON files.*
34
  - Trained on a set of 128 images
 
35
  - 9 repeats
36
  - 3 batch size, 4 epochs
37
  - `(128*9) / 3 * 4` = 1536 steps