circlestone-labs/Anima · Discussions

#53 opened 1 day ago by

gruevy

plz drop the LLM Adapter. It bottlenecks Qwen 0.6B's true potential (NLU & Chinese)

🤯 4

#52 opened 2 days ago by

Schelluwu123

Run times?

#51 opened 4 days ago by

anonanonanon

Can it do image to image?

#50 opened 5 days ago by

koxan17

🔥 [Updated] Style Explorer: 3,000+ Danbooru Artist Comparison

❤️ 👍 13

11

#49 opened 6 days ago by

ThetaCursed

E621 dataset?

👍 9

15

#48 opened 6 days ago by

shiboishi

Using multiple quality tags (NovelAI/Pony/etc) is a good idea

#47 opened 7 days ago by

Shinku

Anima has been trained incredibly well. I wonder when the full version will be released—probably which month?

👀 4

#46 opened 8 days ago by

kexing

Spaces or not, and is 0.6b enough?

😔 1

7

#45 opened 8 days ago by

notafraud

Artist mixing does not work (or at the very least not in the way you'd expect it to)

7

#44 opened 8 days ago by

qtpie

Hey good shit this model trains well. Heres a training config too.

👍 3

#43 opened 9 days ago by

EclipseMist

Is latent upscaling broken for this model?

#42 opened 9 days ago by

Suilibo

NOT WORKING!

🧠 1

5

#41 opened 9 days ago by

VashXP

Very impressive!

👍 2

#40 opened 10 days ago by

kirjolohi

Can I ideally blend monochrome artist style with color artist style?

#39 opened 11 days ago by

jullirin98

How do I use this?

🔥 3

#38 opened 11 days ago by

DMann4214

Consider adding "no text" as a positive quality tag

👍 11

#36 opened 11 days ago by

Caesar69

The training script for the Anima model has already been implemented for sd-scripts

❤️ 13

33

#35 opened 12 days ago by

duongve

Why include ye-pop and deviantart datasets?

👍 1

#34 opened 12 days ago by

amarney

How can I run this model on Google colab?

#33 opened 12 days ago by

888lext

Pony scoring system concerns.

🔥 👍 19

#32 opened 12 days ago by

shiboishi

License concerns

#31 opened 12 days ago by

nonetrix

Open-source/commercial license for Anima?

#30 opened 12 days ago by

JJHaggar

New image upload

#29 opened 12 days ago by

caspiandonavon

Experimental Training code with Full Finetune and LoRA Support Mimicking the features implemented by Tdrussell(for linux or WSL2)

❤️ 🚀 19

#28 opened 13 days ago by

Bluvoll

What Is the Base Model of the Final Model?

👍 3

#27 opened 13 days ago by

nagarago

Tag Data excel?

👍 1

#26 opened 14 days ago by

godlock2006

Prompt outputs failed validation: UNETLoader:

#24 opened 14 days ago by

Andred777

Did someone say... training? :doro: There's a working lora inside this discussion, trust me.

🔥 👍 17

8

#23 opened 14 days ago by

Bluvoll

Unrecognized model?

#22 opened 14 days ago by

Jaycod

FP16_Diagnosis_Report.md

#21 opened 14 days ago by

scvxzf

We're a game design company and...

👀 1

#20 opened 15 days ago by

YellowjacketGames

This Is Amazing So Far

#19 opened 15 days ago by

vennin

Good job

🤯 5

8

#18 opened 15 days ago by

AliceThirty

Wide aspect ratio quality

#17 opened 15 days ago by

notafraud

NVFP4?

➕ 5

#16 opened 15 days ago by

Remile

On my Tesla V100, it's ten times slower than SDXL.

11

#15 opened 15 days ago by

dawn6666

Using “@artist name” causes hand collapse.

👍 2

5

#14 opened 15 days ago by

jullirin98

some feedback

#13 opened 15 days ago by

conki

Very slow inference on AMD Strix Halo

#12 opened 16 days ago by

mikkoph

Possibility of switching to Flux2 VAE

❤️ 🚀 18

#11 opened 16 days ago by

DraconicDragon

Training code?

👍 7

#10 opened 16 days ago by

Darudado

Will the final model be released certainly? + captioning question

👍 18

8

#9 opened 16 days ago by

Norian

is any kind of regional attention supported?

#8 opened 16 days ago by

shivshankar

Can we get an artist list?

#6 opened 16 days ago by

juanml82

pls define "medium resolution"

#5 opened 16 days ago by

Hugs288

Model Use

👍 5

#4 opened 16 days ago by

dearest

ControlNets

❤️ 🔥 5