hhks
/

PonyXL_Notes_Backup

+#PonyXL Notes
+##Get PonyXL here
+https://civitai.com/models/257749?modelVersionId=290640
+Anons have been experimenting with merging PonyXL with other checkpoints, the following autism mixes have been popular, the DPO one has a much better default style out of the box, PonyXL LoRAs do still work on them but they may have a reduced strength.
+https://civitai.com/models/288584?modelVersionId=324619
+##List of Ponyxl LoRAs from /h/
+NEW NOTE 1/24/2024: I've noticed that some LoRA bakers are using different negatives when testing their LoRAs, which can lead to discrepancies when using them. I've mostly noticed that some LoRAs have a significantly reduced effect if you don't have source_pony or some combination of the source_* tags other than source anime in the negatives. This is unfortunate but a byproduct of not having a standardized process for creating the LoRAs. There's another section down below that has more info about similar issues with PonyXL that's worth reading.
+I made a powershell script to try and pull down the LoRAs on this rentry, it will store the last downloaded url for a lora in a history file in case a new or improved one replaces it so you can reuse the script without downloading everything again. You call the script and pass in a destination folder to store the LoRAs. Obviously run at your own risk, I can't prove to you I'm not an asshole or terrible programmer. You'll probably get throttled if you download everything in one go so run it overnight or something. Some of these LoRAs are huge, the whole folder with them all downloaded is ~15 gb.
+UPDATE 1/21/2024 I added a basic management interface so it doesn't just pull down everything. You can pick and choose which loras you want.
+UPDATE 1/22/2024 Added some logic to keep old versions of LoRAs around in case one is updated but you still want to keep the old version around as well.
+https://files.catbox.moe/2kiajb.ps1
+###Random
+https://files.catbox.moe/kwo8w4.safetensors sayo_no_uta a horror game from 2003, you would have to look at the lora metadata to know what to prompt, there's flesh on the walls and stuff, fun to mess with
+https://files.catbox.moe/d9k7nf.safetensors slay_the_princess a visual novel that came out last year that has a sketchy art style, supports 5 of the game's characters: adversary, specter, nightmare, tower, and witch you have to prompt for their features to get them to show up, this file has the characters with their associated tags: https://files.catbox.moe/uprfne.txt, here's some example gens: https://files.catbox.moe/32eskz.png, https://files.catbox.moe/c87hl7.png, https://files.catbox.moe/kp5hbk.png, https://files.catbox.moe/jta8v1.png, https://files.catbox.moe/jsz1rp.png
+###Artists
+These don't require activation triggers unless otherwise stated. Some of these had been previously uploaded to litterbox because they were too large for catbox. Unfortunately the links have expired, I've replaced them with resized versions that are just a couple ranks smaller so hopefully there's no quality loss. These replacement LoRAs have resized on the end of their names. I removed the original ones so people who were lucky enough to download them won't have them overwritten.
+https://files.catbox.moe/icbrs7.safetensors hungry_clicker
+https://files.catbox.moe/dqfgjc.safetensors shexyo
+https://files.catbox.moe/kn021a.safetensors superbusty
+https://files.catbox.moe/ps6xp2.safetensors asanagi requires activation trigger of asanagi in the prompt
+https://files.catbox.moe/0s2xb6.safetensors ushiyama_ame
+https://files.catbox.moe/o97yd0.safetensors jakko
+https://files.catbox.moe/eqcg4d.safetensors kaos_art
+https://files.catbox.moe/9qwl57.safetensors g3mma
+https://files.catbox.moe/nvc1m8.safetensors ndc
+https://files.catbox.moe/kbaz4k.safetensors trogfor
+https://files.catbox.moe/4b97sc.safetensors nephlite
+https://files.catbox.moe/qtpqif.safetensors yd
+https://files.catbox.moe/ecebk5.safetensors rustle
+https://files.catbox.moe/ojbvr4.safetensors afrobull
+https://files.catbox.moe/rqqk5m.safetensors akihiko_yoshida
+https://files.catbox.moe/fauar8.safetensors amazuyu_tatsuki
+https://files.catbox.moe/cuzgnf.safetensors aoi_nagia just using stuff from Elf Mori
+https://files.catbox.moe/9xb1r0.safetensors araneesama
+https://files.catbox.moe/y5xajf.safetensors asashina
+https://files.catbox.moe/yur5xf.safetensors ennuigrl
+https://files.catbox.moe/74oe4g.safetensors enoshim_iki
+https://files.catbox.moe/ntxm1m.safetensors futonpasta The artist from a doujin game circle.
+https://files.catbox.moe/ihmnwq.safetensors kurosu_gatari
+https://files.catbox.moe/navp8q.safetensors kunaboto
+https://files.catbox.moe/imc8ii.safetensors kuroda_akimi Kind of really hard to escape the censor with this one.
+https://files.catbox.moe/mrxxs7.safetensors ryoji_ohara Main artist for the Final Fantasy Tactics War of the Visions.
+https://files.catbox.moe/7ozz68.safetensors haganeF
+https://files.catbox.moe/ck31b2.safetensors gentle_sasaki
+https://files.catbox.moe/gr5pdx.safetensors gaptax will likely display issues at some point due to small dataset
+https://files.catbox.moe/986c8k.safetensors Gamukami
+https://files.catbox.moe/0wssp8.safetensors Fishine
+https://files.catbox.moe/xbeu1x.safetensors Deepvalley will likely display issues at some point due to small dataset
+https://files.catbox.moe/9h5vos.safetensors shinonome_kazuhiko
+https://files.catbox.moe/tevomb.safetensors chocochip
+https://files.catbox.moe/fbwz5k.safetensors piromizu
+https://files.catbox.moe/odlshx.safetensors piromizu_2
+https://files.catbox.moe/4xmzn0.safetensors porkstudio
+https://files.catbox.moe/0ebohp.safetensors starworks
+https://files.catbox.moe/q1vapd.safetensors suimitsutou_koubou
+https://files.catbox.moe/4gd1nk.safetensors hakika
+https://files.catbox.moe/c54e0t.safetensors homare
+https://files.catbox.moe/pooszm.safetensors komori_kei
+https://files.catbox.moe/cysexx.safetensors MetalOwl
+https://files.catbox.moe/wy3knx.safetensors Methonium
+https://files.catbox.moe/4i8oae.safetensors Oda_Non
+https://files.catbox.moe/b5eyg8.safetensors ratatatat74
+https://files.catbox.moe/frxmlo.safetensors Maron_Maron requires activation trigger "by maron maron"
+https://files.catbox.moe/m764gt.safetensors Tsubasa_tsubasa_resized
+https://files.catbox.moe/g42ahp.safetensors tenk
+https://files.catbox.moe/cos32m.safetensors tianliang_duohe_fangdongye_resized
+https://files.catbox.moe/vog70p.safetensors gud0c
+https://files.catbox.moe/ccz6sl.safetensors mo_(kireinamo)
+https://files.catbox.moe/1dsnon.safetensors konpeto
+https://files.catbox.moe/8hg3yk.safetensors melon22 It's pretty versatile, but to get the true melon look you have to add some helper tags. Try huge breasts, (muscular female:0.5), and neg muscular arms.
+https://files.catbox.moe/qj7xl0.safetensors pepper0
+https://files.catbox.moe/zi3lzf.safetensors mikozin
+https://files.catbox.moe/f0y8ew.safetensors sake_kemosalmon
+https://files.catbox.moe/wcez69.safetensors yuta_agc_kids_gamera
+https://files.catbox.moe/2wanow.safetensors BlueTheBone
+https://files.catbox.moe/4dhrck.safetensors George_Kamitani
+https://files.catbox.moe/4pr3h4.safetensors Gerupin
+https://files.catbox.moe/krtbno.safetensors Harada
+https://files.catbox.moe/1ndvd4.safetensors Himukai_Yuuji
+https://files.catbox.moe/pagud0.safetensors Takeda_Hiromitsu
+https://files.catbox.moe/n9pe8s.safetensors Kuuchuu_Yousai
+https://files.catbox.moe/58qele.safetensors Mekabu_Aji_MAX
+https://files.catbox.moe/g29yo2.safetensors Nyamota
+https://files.catbox.moe/8p34wl.safetensors Urushima
+https://files.catbox.moe/i013hk.safetensors Warabimochi
+https://files.catbox.moe/aqm2v0.safetensors Xration
+https://files.catbox.moe/q6f33q.safetensors Yageshi_Nan
+https://files.catbox.moe/caabk4.safetensors Zankuro
+https://files.catbox.moe/qybjzv.safetensors cyberboi
+https://files.catbox.moe/vw47lk.safetensors gorani
+https://files.catbox.moe/v8qsfh.safetensors mojihe
+https://files.catbox.moe/m7sa64.safetensors owner
+https://files.catbox.moe/rhlwp5.safetensors woohyoot
+https://files.catbox.moe/rhlwp5.safetensors yeehawt0wn
+https://files.catbox.moe/bj5m7b.safetensors nawakena_resized example gen: https://files.catbox.moe/26uxu5.png
+https://files.catbox.moe/opwfd2.safetensors slush
+https://files.catbox.moe/2ei4zl.safetensors aoin_hatsu
+https://files.catbox.moe/cd6exz.safetensors andava
+https://files.catbox.moe/j10wt5.safetensors source_filmmaker
+https://files.catbox.moe/o1fmsv.safetensors namako_daibakuhatsu
+https://files.catbox.moe/800fnr.safetensors ishikei you can maybe put source_anime in the positive prompt to make it look more like his old art but as far as i can tell that just happens no matter what
+https://files.catbox.moe/v95tlf.safetensors murata_range works well on both pony and the autism_pony mix
+https://files.catbox.moe/3aewnu.safetensors rhasta you probably need to use source_anime in the positive prompt and/or nose in the negative prompt to stop the model from deciding to add realistic noses
+https://files.catbox.moe/kh4goe.safetensors laserflip
+https://files.catbox.moe/iq5u22.safetensors ibukichi
+https://files.catbox.moe/29oqmg.safetensors kurono_mitsuki
+https://files.catbox.moe/mj8sls.safetensors tokkyu
+https://files.catbox.moe/s5n8ul.safetensors moshimoshibe
+https://files.catbox.moe/amb4x8.safetensors shuz_ requires activation prompt of shuz and helps to put 3d in the negatives
+https://files.catbox.moe/a0rftx.safetensors kazutake_hazano
+https://files.catbox.moe/bxoljw.safetensors fujou_joshi
+https://files.catbox.moe/cbrvwe.safetensors millefeuille helps to put monochrome and greyscale in the negatives
+https://files.catbox.moe/sc7pis.safetensors tottotenero add text and speech bubble to the negative and maybe some quality tags to the positive prompt and it should stop it from generating text
+https://files.catbox.moe/odmswn.safetensors fkey
+These are resized version of the LoRAs below so they fit on catbox and will thus download automatically with the powershell script, I didn't test them after resizing so can't comment on quality loss.
+https://files.catbox.moe/jtgi98.safetensors kezime
+https://files.catbox.moe/pr6abi.safetensors menma
+https://files.catbox.moe/pzrhxx.safetensors nullmeta
+https://files.catbox.moe/mlziu8.safetensors Megami_Magazine
+https://files.catbox.moe/oqhdry.safetensors nyantcha
+https://files.catbox.moe/8i49kc.safetensors shirow
+https://files.catbox.moe/rp7cbf.safetensors onono_imoko
+https://files.catbox.moe/5pw11p.safetensors error1980
+https://files.catbox.moe/xdmcjd.safetensors non_(z-art)
+https://files.catbox.moe/2cl9sq.safetensors tsukumizu_yuu
+https://files.catbox.moe/adligx.safetensors dratvan
+https://files.catbox.moe/pmv83f.safetensors poper
+https://files.catbox.moe/593c8m.safetensors nns
+The following will not autodownload using the script, you'll have to manually do it, can't be assed to figure out how to download from megaupload or managing zip files.
+https://mega.nz/file/xHt0kAQQ#lyOUMAOfm5MP60uNhCdT87Eaw8R6jB-oMpHbF_XBxEM mena
+https://mega.nz/file/pbszzZBA#WH08IpwvquRcueGie40IznEDMnV-jIsl-vLBaVfFWXg nullmeta (kokaki mumose)
+https://mega.nz/folder/83gQXTLT#mrjDP3w_OkxR0ujuVliesw kezime
+https://files.catbox.moe/b4wk0u.zip Megami_Magazine flat anime style
+https://files.catbox.moe/zxvyki.7z nyantcha
+https://mega.nz/file/V9V13KaQ#oeIASCTJrcK7CMJpFLHyqBWLlm0sGkFnvG9in8vKs9A slush
+https://mega.nz/folder/83gQXTLT#mrjDP3w_OkxR0ujuVliesw/folder/xuRSSbDK shirow,onono_imoko,kezime
+https://mega.nz/folder/RG81USjC#0T7bFau2u6bDCSodJJWX4w nullmeta,non_(z-art),tsukumizu_yuu,error1980,nns
+###Characters
+https://files.catbox.moe/1wdl00.safetensors character_momoka helps to prompt features like hair, clothes
+https://files.catbox.moe/f5ssgz.safetensors character_tachibana_arisu helps to prompt features like hair, clothes
+https://files.catbox.moe/9zcayd.safetensors character_matoba_risa helps to prompt features like hair, clothes
+https://files.catbox.moe/djnlks.safetensors character_amy_sorel example prompt image here: https://files.catbox.moe/qymyf8.png, trigger is amy sorel, outfits are tagged as sc4 and sc6
+https://files.catbox.moe/xow8kx.safetensors character_maryjane Prompt is maryjane, though it works better if you prompt for other stuff as well. gradient hair, tentacle hair, head fins, fangs, long tongue, etc
+https://files.catbox.moe/ccbfrs.safetensors character_neopolitan_rwby activation seems to be neo_politan, example gen: https://files.catbox.moe/m77z3a.png
+https://files.catbox.moe/0fmlta.safetensors character_naoto_shirogane looks like Shirogane Naoto is the activation trigger, example prompt image: https://files.catbox.moe/hf88vs.png
+https://files.catbox.moe/vsky78.safetensors character_shinobu_jackobs example prompt image: https://files.catbox.moe/jale97.png, activation trigger is shinobu jackobs
+https://files.catbox.moe/z8j9aq.safetensors character_roll_exe
+https://files.catbox.moe/ifr6a5.safetensors character_mayl_sakurai activation trigger is maylsakurai, can also prompt for the following 'blue vest', 'green sleeves', 'pink skirt', 'black thighhighs', 'hair ornament'
+https://files.catbox.moe/e9ult9.safetensors character_scarlet_ff7 activation trigger is scarlet_\(ff7\), also helps to prompt asymmetrical hair, blonde_hair, red dress if you want her in her typical red dress, looks best with 3d in the prompt due to the limited amount of images out there, once you put 3d in the negatives it starts to look a lot less like her
+https://files.catbox.moe/mxnwc6.safetensors character_fern
+##Innate artists/styles
+List of tags innate to ponyxl that change the style significantly, usually artists but sometimes other strange behavior some of these could just be getting activated by character names in the artist name or just misfires in the AI network ¯\_(ツ)_/¯
+asura \(asurauser\)
+mirai hikari
+cutesexyrobutts
+bb \(baalbuddy\) (questionable quality)
+dd \(ijigendd\) (not sure if it actually matches the real artist)
+hara \(harayutaka\)
+hews
+boris \(noborhys\)
+m-da s-tarou
+tsukishiro_saika
+tsukudani_(coke-buta) (strange behavior with hair)
+agawa_ryou (seems pretty faithful but lips are always really full)
+sumiyao \(amam\)
+neocoill
+tomose shunsaku,
+null \(nyanpyoun\),
+horosuke
+chanta \(ayatakaoisii\), (hard to tell if it's faithful)
+merunyaa (seems to be the character not really the style)
+fumihiko \(fu mihi ko\) (always drenched in sweat for some reason)
+aaaa \(quad-a\) (I like this one!)
+cle masahiro (everything has rim lighting)
+kasumi \(skchkko\)
+murakami suigun (not sure if actually the artist, pretty strange eyes)
+fumio \(rsqkr\)
+yomu \(sgt epper\)
+homare \(fool's art\)
+happoubi jin (seems to be slight? not sure)
+asamura hiori (doens't seem faithful but looks interesting!)
+moisture \(chichi\) (seems faithful)
+dandon_fuga
+kase daiki
+taka \(taka wo kukuru\)
+ririko_\(zhuoyandesailaer\)
+mo_\(kireinamo\)
+akinbo_\(hyouka_fuyou\)
+aztodio
+kaisen_chuui
+sousouman
+bee_\(deadflow\)
+sano_toshihide
+mitsumi_misato
+personal_ami
+donguri_suzume
+lambda_\(kusowarota\)
+jack_dempa
+lm_\(legoman\)
+minakami_\(flyingman555\)
+jellytits-7
+amasora_taichi
+em_\(totsuzen_no_hakike\)
+kurokoshou_\(emuburemu123\)
+niliu_chahui
+kawakami_rokkaku
+miyase_mahiro
+momo_no_sukebe
+minato_ojitan
+joy_ride
+osiimi
+mosho
+fuya_\(tempupupu\)
+jabara_tornado
+konno_tohiro
+23_\(real_xxiii\)
+kashu_\(hizake\)
+mizuryu_kei
+kairunoburogu
+mizuumi_\(bb\)
+nagase_haruhito
+nottytiffy
+akashio_\(loli_ace\)
+liang_xing
+vahn_yourdoom
+nakamura_regura
+aster_crowley
+cian_yo
+hekiga_\(freelot\)
+tsukiyo_\(skymint\)
+nyong_nyong
+tony_guisado
+tebi_\(tbd11\)
+neoartcore
+yoi_naosuke
+damao_yu
+masami_chie
+from here will try to classify them better, too lazy to revisit past ones
+strong and accurate (may need strength tweaking)
+tony_taka
+dross
+hua-j
+sparrow
+firolian
+paloma_piquet
+akano_murasaki
+Kawamoto Toshihiro
+blade_(galaxist)
+bittenhard
+stealth-brock, (also maybe shenanimation, alt)
+incase
+sunna \(nilsunna\)
+wagashi \(dagashiya\)
+sagiri
+melkor_mancin
+alp
+elf \(stroll_in_the_woods\)
+tomu \(tomubobu\)
+kame \(kamepan44231\)
+cyancapsule
+kelvin hiu
+neisan
+pija \(pianiishimo\),
+strong and doesn't look anything like the artist imo but still possibly useful through varying strength and combining with other artists
+kelvin-trainerk (maybe something to do with kelvin above?)
+geo_\(yukishitadou\)
+fadingz
+sima_naoteng
+nyororiso_\(muyaa\)
+seems to induce a change but I could be hallucinating or it could
+be just due to the randomness of the model
+nemunemu_\(candy_paddle\)
+kurokaze_no_sora
+kapuru_0410
+iro_ame_\(amewaagada\)
+hisakawa_chin
+sunga2usagi
+shirokuma_\(nankyoku\)
+kokura_masashi
+hinaki_\(hinaki_0102\)
+haru_\(hiyori-kohal\)
+hanamasa_ono
+amano_kozue
+ziro_\(zirorong\)
+shiromoru_\(yozakura_rety\)
+sakura_hanpen
+nyonyonba_tarou
+miyuki_\(miyuki0529\)
+ichirino_minagi
+shimashiro_itsuki
+nijihashi_sora
+mugichoko_\(mugi_no_choko\)
+kamimiya
+haneramu
+nakamura_rohane
+oimo_\(imoyoukan\)
+mugi_\(banban53\)
+mana_kakkowarai
+last tried was yanyan_\(shinken_gomi\)
+##Innate Characters
+An anon went through the trouble of compiling a  list of characters known to the model, these are not comprehensive it seems to know hundreds if not thousands of characters. Any popular female character from a popular franchise has a good chance of being known to the model, movies, video games, and tv shows are a good source.
+Females: https://files.catbox.moe/oklpz8.txt
+Males: https://files.catbox.moe/1lptzn.txt
+Female Pokemon: https://files.catbox.moe/sljtm8.txt
+Male Pokemon: https://files.catbox.moe/ka331j.txt
+Western Characters: https://files.catbox.moe/15bayo.txt
+More Western Characters: https://files.catbox.moe/bhjnpb.txt
+##Tricks with the model for more anime style
+There's some evidence that the model has a bit of a bias towards western style art, particularly when using the score tags (it is made by furries after all). It might be beneficial to tag your image sets with the score tags, the laziest way would be to tag every image with score_9, source_anime, this may affect the overall "quality" of images the lora generates since some of the knowledge in the model about "high quality" might be overwritten but it will look more like the artist's style.
+You can also help nudge generations with the model more towards anime style by using "source_cartoon, source_furry, source_pony, sketch, painting, monochrome" in the negative prompt when generating images. Some of the artists with more subtle art styles tend to have more success with this in my experience. The bias is a bit less evident in the lower scores so if a LoRA has it's images tagged with score_9 a prompt like "source_anime, score_9, score_6_up, score_5_up, score_4_up" might get better results. Unfortunately this also makes your LoRA harder to use, you'd have to tell people to use it this way.
+Other tricks tried were to include all the score tags in the LoRA's training set on every image which didn't have much success or only using the source_anime tag which also didn't seem to influence the LoRA's effectiveness much. I haven't tried actually using the score tags as intended since that takes more effort than I'm willing to put in for a LoRA for the time being but that may net the best results.
+##LoRA Training Update 1/30/2024
+I've been messing around with Lycoris locon, which needs inputs for conv network rank and alpha which I generally saw recommendations to set it to the network rank and alpha of the model (and to not exceed 64). Not sure how easy it is to use with the UI frontends but here's what the commandline looks like for kohya-ss. As always any criticism is welcome on /hdg/ I've actually gotten some good feedback on there.
+ --network_module = networks.lora
+ --train_data_dir = "$IMAGES_PATH$"
+ --output_dir = "$OUTPUT_FOLDER$"
+ --output_name = "$LORA_NAME$"
+ --pretrained_model_name_or_path = "$PATH_TO_PONYXL_CHECKPOINT$"
+ --max_train_epochs = $NUMBER_EPOCHS$
+ --train_batch_size = 12
+ --resolution = "1024,1024"
+ --save_every_n_epochs = 1
+ --save_last_n_epochs = 999
+ --learning_rate = 1
+ --lr_scheduler = cosine
+ --lr_warmup_steps = 0
+ --network_dim = $NETWORK_RANK$
+ --network_alpha = $NETWORK_ALPHA$
+ --seed = 1056283418
+ --keep_tokens = 0
+ --gradient_checkpointing
+ --max_data_loader_n_workers = 16
+ --mixed_precision = bf16
+ --save_precision = bf16
+ --network_module = lycoris.kohya
+ --network_args "preset = unet-transformer-only" "conv_dim = $NETWORK_RANK$" "conv_alpha = $NETWORK_ALPHA$" "algo = locon"
+ --optimizer_type = "prodigy"
+ --optimizer_args "decouple = True" "weight_decay = 0.01" "d_coef = 0.8" "use_bias_correction = True" "safeguard_warmup = False" "betas = 0.9,0.99"
+ --caption_extension = ".txt"
+ --prior_loss_weight = 1
+ --cache_latents
+ --cache_latents_to_disk
+ --xformers
+ --save_model_as = safetensors
+ --enable_bucket
+ --min_bucket_reso = 512
+ --max_bucket_reso = 2048
+ --bucket_reso_steps = 256
+##Old LoRA training settings
+These are my old LoRA training settings with prodigy, didn't really have any major issues with it, keeping it around for reference.
+ --network_module = networks.lora
+ --train_data_dir = "$IMAGES_PATH$"
+ --output_dir = "$OUTPUT_FOLDER$"
+ --output_name = "$LORA_NAME$"
+ --pretrained_model_name_or_path = "$PATH_TO_PONYXL_CHECKPOINT$"
+ --max_train_epochs = $NUMBER_EPOCHS$
+ --train_batch_size = 2
+ --resolution = "1024,1024"
+ --save_every_n_epochs = 1
+ --save_last_n_epochs = 999
+ --learning_rate = 1
+ --lr_scheduler = cosine
+ --lr_warmup_steps = 0
+ --network_dim = 64
+ --seed = 1055217506
+ --keep_tokens = 0
+ --gradient_checkpointing
+ --max_data_loader_n_workers = 8
+ --mixed_precision = bf16
+ --save_precision = bf16
+ --optimizer_type = "prodigy"
+ --optimizer_args "decouple = True" "weight_decay = 0.01" "d_coef = 0.8" "use_bias_correction = True" "safeguard_warmup = False" "betas = 0.9,0.99"
+ --caption_extension = ".txt"
+ --prior_loss_weight = 1
+ --enable_bucket
+ --min_bucket_reso = 512
+ --max_bucket_reso = 2048
+ --bucket_reso_steps=256
+ --xformers
+ --save_model_as = safetensors
+ --cache_latents
+ --cache_latents_to_disk
+ --network_train_unet_only
+ --cache_text_encoder_outputs