Phospy commited on
Commit
c38a8bd
·
1 Parent(s): 7759afc

Delete mega_file

Browse files
mega_file/Kuon_2800.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:27e0f9c0706a83c03644f27c647096ede1a6074045278d570192116a3d2a3996
3
- size 151074558
 
 
 
 
mega_file/Noa.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:090e5c5cffe31728dbcf89d4800d55091d8a985296d9f5f7d23544b9c616ca63
3
- size 18951616
 
 
 
 
mega_file/Pokemon - Mallow v1 (2).safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:8a74c22d1c2616218a7a8df4d4de577cf87570e715c8d9adc5af5301ef7734a6
3
- size 151074560
 
 
 
 
mega_file/concept_pronebone.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:ef26bc9cf01d150166e3385c54b8987de2b722b923afa95b76e603ef3419e526
3
- size 151074550
 
 
 
 
mega_file/concept_pronebone.txt DELETED
@@ -1,22 +0,0 @@
1
- trained on 175 images from gelbooru between 3 folders
2
- 5_prone bone, 95 images
3
- 8_from side, 55 images
4
- 15_head push, 25 images
5
- 128 dim
6
-
7
- batch size 4
8
- epochs 8
9
-
10
- TAGS:
11
-
12
- pronebone, prone bone, boy on top, deep penetration
13
-
14
- for side view, add: from side
15
- for hand on head: hand on another's head, rough sex, head push, held down
16
-
17
- a good place to start is:
18
- pronebone, boy on top, 1boy, 1girl, hetero, sex
19
-
20
- at 1.1 weight
21
-
22
- example files are also in this folder, mostly tested using aom2-nsfw
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
mega_file/japanese clothes v2.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3b4d6c9cde2748019249503f66af3dc5e71f4744259b37d26ffbbb3b27c45b59
3
- size 151073665
 
 
 
 
mega_file/japanese clothes.txt DELETED
@@ -1,22 +0,0 @@
1
- This training file is for characters who cannot wear kimono or yukata.
2
-
3
- Working - furisode , bath yukata , short yukuta uchikake[wedding] , yukata[working] , short kimono[working] ,
4
- Under training -
5
- Not training - layered kimono , kimono skirt , hiki furisode[low data] ,
6
-
7
- How to use
8
- - LoRA str = 1
9
- - LoRA Character str = 0.5-1.0
10
-
11
- --- v2 ---
12
- yukata = short yukata / negative prompt : legs,
13
- uchikake = uchikake, hood,
14
- --- v1 ---
15
- Furisode prompt = furisode, kimono, long sleeves, wide sleeves, obi, sash, fur-trimmed kimono, fur trim, fur collar, fur scarf,
16
- bath yukata = bath yukata, / negative prompt : obi, sash,
17
- short yukata = short yukata, yukata, legs,
18
-
19
- Data image = 4 folder 330 image / 5-9 repeats / 2 epochs / 1 batch
20
- learning_rate = 1e-4
21
- network_dim=128
22
- scheduler="cosine_with_restarts"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
mega_file/kuon.txt DELETED
@@ -1,3 +0,0 @@
1
- Reacts to "Kuon, animal ears" Adult version was trained on for a bit aswell, "Adult Kuon" is the #1 tag on it. It might seep into the model anyway if you try to change clothes.
2
-
3
- Barely tested, 1400+ should be fine, haven't checked how easy is it to change clothing etc
 
 
 
 
mega_file/michiru-v4-epoch03.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a6c8a8dd951eeeadb47fce164227e58564c6187f10cbd94efde1af51a0ee8471
3
- size 151074584
 
 
 
 
mega_file/michiru.txt DELETED
@@ -1,31 +0,0 @@
1
- # Chidori Michiru (Blue Archive)
2
- ## Usage
3
- Use any or all of these tags to summon Koharu:
4
- `michiru, 1girl, halo, yellow eyes, grey hair, small breasts`
5
- Tag pruning means you don't need to prompt for much other than `michiru` to get a pretty good result -- eyes and hair are optional but can help correct occasional mistakes.
6
- The AI really tries to do her ninja `kuji-in` but unsurprisingly fucks it up more often than not, holding up the wrong number of fingers or just generating messed up hands.
7
- The tail is a bit iffy -- I tagged it as `tail` but prompting for that gets a cat tail half the time. Try `raccoon tail`.
8
-
9
- For her normal outfit:
10
- `school uniform, blue skirt, black pantyhose, floral print, black scarf, bridal gauntlets`
11
- I tagged images with `sarashi` even when it was only partially visible under her uniform.
12
-
13
- Unlike Koharu's LoRA I was a little more specific with clothing colors when tagging. Skirts are always tagged `blue skirt`, scarf is always `black scarf` etc. The hope was that it would make it possible to get her normal clothing in a different color, and to avoid overfitting `skirt` and `scarf` into the exact ones she normally wears. It sorta works but if you want a`red skirt` you have to emphasize it and negative prompt `blue skirt`.
14
-
15
- Weights from 0.8 - 1.05 should work well, it's perhaps slightly overfitted.
16
-
17
- ## Training
18
- *All parameters are provided in the accompanying JSON files.*
19
- - Trained on a curated set of 88 images repeated 10 times.
20
- - Dataset included a mixture of SFW and NSFW.
21
- - This dataset was smaller than Koharu's, but I maintained the same number of steps.
22
- - Initially tagged with WD1.4, then performed heavy pruning and editing.
23
- - Removed as many inaccurate tags as possible
24
- - Made sure important traits were present and consitently described, and traits like `halo` were consistent with actual visibility
25
- - Pruned redundant tags and simplified outfits so that they were always tagged with the same handful of tags
26
- - Added camera angles and image composition hints
27
- - Added a few facial expressions
28
- - Different learning rate than usual.
29
- - 5e-5 text encoder (same as Koharu's, but typically 1e-5 ~ 2e-5)
30
- - 3e-4 UNet (typically one order of magnitude faster than text)
31
- - Trained without VAE.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
mega_file/noa.txt DELETED
@@ -1,8 +0,0 @@
1
- Recommended prompt = halo, white shirt, jacket two-sided jacket, open jacket, necktie, white hair, purple eyes, white skirt,
2
- LoRA str = 1
3
-
4
- learning_rate = 5e-5
5
- network_dim=16
6
- scheduler="cosine_with_restarts"
7
-
8
- Training = 98 file image / 5 Repeats / idk epochs
 
 
 
 
 
 
 
 
 
mega_file/pee+lactation.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:e4e7d5523df604017c043377c41066a859554deeda550581c8a6de42f70a63e0
3
- size 151073665
 
 
 
 
mega_file/pee_lactation.txt DELETED
@@ -1,10 +0,0 @@
1
- prompt = pee, peeing,
2
- lactation = lactation, nipples, or lactation, lactation through clothes,
3
-
4
- LoRA str = 1
5
-
6
- learning_rate = 5e-5
7
- network_dim=128
8
- scheduler="cosine_with_restarts"
9
-
10
- Training = 202 file image / 7 Repeats / 2 epochs
 
 
 
 
 
 
 
 
 
 
 
mega_file/pregnant.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:d6e21282969b51f181b22e3b0ac0273931be0dd99d47f0fe43a47401b0ff55b6
3
- size 151073665
 
 
 
 
mega_file/pregnant.txt DELETED
@@ -1,16 +0,0 @@
1
- prompt = pregnant,
2
- multiple pregnant = multiple pregnant,
3
-
4
- If increasing (pregnant:1.0-1.5) will increase belly size or reduce to pregnant:0.8 if you don't want to be big belly
5
- Sometimes larger pregnant, prompt with huge breasts,
6
-
7
- in folder dim 16-64 Comparison can be seen in image.
8
- It is recommended to use 64 or 128 for better pregnancy quality
9
-
10
- LoRA str = 0.6-1
11
-
12
- learning_rate = 9e-5
13
- network_dim=128
14
- scheduler="cosine_with_restarts"
15
-
16
- Training = 151 file image / 5-7 Repeats / 2 epochs
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
mega_file/quasarcake-last.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0f18259435438b77cde73102069bdcf0bbe57c91f90b8ca0b3963557d366e978
3
- size 151073665
 
 
 
 
mega_file/virgin killer sweater.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:db961a4a5708bc799e5428248f23a3a95712c4c28119ba69040f00aacda696d5
3
- size 151073665
 
 
 
 
mega_file/virgin.txt DELETED
File without changes