adds retrained Sora LoRA

README.md

@@ -6,16 +6,54 @@ license: mit
 
 Here you will find the various LoRAs I've trained, typically of Blue Archive characters.
 
-## 🚧 IMPORTANT 🚧
-
-I haven't moved all of my LoRAs over from Mega.  Please see the [Mega.co.nz repository](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA) if a LoRA is missing from here.
-
 ## Blue Archive
 ブルーアーカイブ / 블루 아카이브
 
+### Atsuko
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Chise
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Hibiki
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Hina
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Iroha
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Izuna
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Kokona
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Mari
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Michiru
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Miyako (WIP)
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Mutsuki
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Natsu
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
+### Resia
+[Available on old Mega.co.nz repository.](https://mega.nz/folder/SqYwQTRI#GN2SmGTBsV6S4q-L-V4VeA)
+
 ### Shizuko
 Kawawa Shizuko / 河和シズコ / 카와와 시즈코
 
 [![Shizuko](shizuko/chara-shizuko.png)](https://huggingface.co/khanon/lora-training/blob/main/shizuko/README.md)
 
+### Sora
+Sora / ソラ (ブルーアーカイブ) / 소라 (블루 아카이브)
 
+[![Sora](sora/chara-sora-v3.png)](https://huggingface.co/khanon/lora-training/blob/main/sora/README.md)

sora/README.md

@@ -1,30 +1,51 @@
-# Sora (Blue Archive NPC)
-Decided to try completely changing things up after seeing an anon's suggestions on /hdg/.  Used batch size 12, lower resolution, and a high learning rate.  Can't tell if it's any better or worse quality-wise than my usual LoRAs, but it was a lot faster to train.
+# Sora (Blue Archive)
+ソラ (ブルーアーカイブ) / 소라 (블루 아카이브)
 
-Be forewardned that this LoRA is prone to generating NSFW results since Sora an unusually high proportion of very lewd art.
+Be forewarned that this LoRA is prone to generating NSFW results since Sora an unusually high proportion of extremely lewd art.  Use `nsfw, naked apron` etc in the negative prompt to prevent this.
+
+[**Download here.**](https://huggingface.co/khanon/lora-training/blob/main/sora/chara-sora-v3.safetensors)
+
+## Table of Contents
+- [Preview](#preview)
+- [Usage](#usage)
+- [Training](#training)
+- [Revisions](#revisions)
+
+## Preview
+
+![Sora portrait](chara-sora-v3.png)
+![Sora](example-001-v3-defmix1.png)
+![Sora fallen](example-002-v3-aom2nmm.png)
 
 ## Usage
-Didn't do much tag pruning so you have to prompt a lot.
 
 Use any or all of the following tags to summon Sora: `sora, halo, mini wings, white wings, two side up, blonde hair, forehead`
 
-For her usual outfit, `blue apron, white shirt` is typically enough.  You can also add `off-shoulder, strap slip, bowtie, short sleeves` if desired.
+For her usual outfit: `blue apron, white shirt, (black pleated skirt:0.8)`
+- If needed, you can also add `off-shoulder, strap slip, bowtie, short sleeves`
+- Sora's skirt is not visible in most training art.  Consider adding `black pleated skirt` and negative prompt `miniskirt, pencil skirt` for more accuracy.
 
-Sometimes her wings appear too high; `low wings` can help with this.  Her wings are also usually a bit too big but that's kind of hard to control.
+Sometimes her wings appear too high; `low wings` can help with this.
 
-Weight 1 seems to work fine.
+Weight 1 is suggested.
 
 ## Training
-*All parameters are provided in the accompanying JSON files.*
-- Trained on a set of 128 images, repeated 25 times, 3 epochs (128 images * 25 repeats / 12 batch size * 3 epochs = 800 steps)
-  - Dataset included a mixture of SFW and NSFW.
-- Batch size 12 is up from usual of 3
-- Resolution was 512, down from usual of 832
-- General learning rate is 1.2e-3, significantly up from the Kohya default (1e-6 I believe)
-  - My understanding is that when this learning rate is set at the same time as the text encoder and unet rates, it is used only as one of the inputs to the AdamW optimizer, which adjusts the learning rate as necessary.  So even though this value is high, it is not running at this value constantly.
-- Text encoder learning rate of 1.5e-5
-- Unet learning rate of 1.5e-4
+*Exact parameters are provided in the accompanying JSON files.*
+- Trained on a set of 128 images; 88 swimsuit, 37 normal.
+  - Datset included a mixture of SFW/NSFW. Sora has more NSFW than most students.
+  - 9 repeats
+  - 3 batch size, 4 epochs
+  - `(128*9) / 3 * 4` = 1536 steps
+- 768x768 training resolution
 - `constant_with_warmup` scheduler instead of `cosine`
 - Initially tagged with WD1.4 Convnextv2 model. Tags minimally pruned/edited.
-- Used network_dimension 128 (same as usual) and network_alpha 64
+- Used network_dimension 128 (same as usual) / network alpha 128 (default)
 - Trained without VAE.
+
+## Revisions
+- v3 (2023-02-06)
+  - Completely re-trained to correct issue where caption files were not being loaded correctly.  New version should have more flexibility but requires additional outfit prompting.
+- v2b (2023-02-01)
+  - Initial release.
+  - **Issue:** caption files were completely ignored during training. As a result, prompting "sora" grants a very strong effect, but the LoRA is highly overfit.
+  - [Download available on Mega.co.nz repository.](https://mega.nz/folder/yjwGiIBD#w79CxW2-Cj66QmObg0zOdg)

sora/chara-sora-v3.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14f8128b46cba1bea83f56316434ea70dc5d3fd17547d6328ff4b30c6309f4b4
+size 151120215

sora/lora_chara_sora_v3_128i9r.json

@@ -0,0 +1,54 @@
+{
+  "pretrained_model_name_or_path": "G:/sd/repo/models/Stable-diffusion/nai-animefull-final-pruned.safetensors",
+  "v2": false,
+  "v_parameterization": false,
+  "logging_dir": "",
+  "train_data_dir": "G:/sd/training/datasets/sora/dataset",
+  "reg_data_dir": "",
+  "output_dir": "G:/sd/lora/trained/chara/sora",
+  "max_resolution": "768,768",
+  "learning_rate": "1e-5",
+  "lr_scheduler": "constant_with_warmup",
+  "lr_warmup": "5",
+  "train_batch_size": 3,
+  "epoch": "4",
+  "save_every_n_epochs": "",
+  "mixed_precision": "fp16",
+  "save_precision": "fp16",
+  "seed": "31337",
+  "num_cpu_threads_per_process": 32,
+  "cache_latents": true,
+  "caption_extension": ".txt",
+  "enable_bucket": true,
+  "gradient_checkpointing": false,
+  "full_fp16": false,
+  "no_token_padding": false,
+  "stop_text_encoder_training": 0,
+  "use_8bit_adam": true,
+  "xformers": true,
+  "save_model_as": "safetensors",
+  "shuffle_caption": true,
+  "save_state": false,
+  "resume": "",
+  "prior_loss_weight": 1.0,
+  "text_encoder_lr": "1.5e-5",
+  "unet_lr": "1.5e-4",
+  "network_dim": 128,
+  "lora_network_weights": "",
+  "color_aug": false,
+  "flip_aug": false,
+  "clip_skip": 2,
+  "gradient_accumulation_steps": 1.0,
+  "mem_eff_attn": false,
+  "output_name": "chara-sora-v3",
+  "model_list": "",
+  "max_token_length": "150",
+  "max_train_epochs": "",
+  "max_data_loader_n_workers": "",
+  "network_alpha": 128,
+  "training_comment": "128 images * 9 repeats / 3 batch size * 4 epochs = 1536 steps",
+  "keep_tokens": 2,
+  "lr_scheduler_num_cycles": "",
+  "lr_scheduler_power": "",
+  "persistent_data_loader_workers": true
+}