huilo huilo

[huilo huilo @marcoc2]

@ostrisai Omg, I just learned how to train loras for ideogram...

[huilo huilo @marcoc2]

@ostrisai @ideogram_ai Yes, please 🙏
Does it need jsons-prompt?

[huilo huilo @marcoc2]

@bdsqlsz Does it need to be trained with jsons?

[huilo huilo @marcoc2]

Check out the new Deep Synth album "Seno Alpha" distributed by @DistroKid and live on Spotify! https://t.co/HzI773lBw8

[huilo huilo @marcoc2]

@ostrisai 🥹🙏

[huilo huilo @marcoc2]

Hate this kind of benchmarking

[Simon Willison @simonw]

Shocking result on my pelican benchmark this morning, I got a better pelican from a 21GB local Qwen3.6-35B-A3B running on my laptop than I did from the new Opus 4.7!

Qwen on the left, Opus on the right

[huilo huilo @marcoc2]

@ostrisai also, I had training lokrs for the previous versions and it was already great https://t.co/RkTafUHLqa

[huilo huilo @marcoc2]

@ostrisai I wonder what band could be this one...

[huilo huilo @marcoc2]

@bdsqlsz Thank you. The original repository is giving me OOM for 24gb when trying to train loras with XL

[huilo huilo @marcoc2]

@wildmindai you could add this workflow as json file on the repo

[huilo huilo @marcoc2]

@DatTwiterGuy @HuggingModels do you have links?

[huilo huilo @marcoc2]

@junmingong @bdsqlsz Did it test with multilanguage?

[huilo huilo @marcoc2]

@ostrisai @bfl_ml 4B and 9B aren't compatible, right?

[huilo huilo @marcoc2]

@LucasPereiraWy @ostrisai @bfl_ml 4B and 9B aren't compatible, right?

[huilo huilo @marcoc2]

@Alibaba_Qwen Qwen-Image is awesome, but please, update it with better textures for realism 🙏🙏

[huilo huilo @marcoc2]

everything is explained now

[huilo huilo @marcoc2]

everything is explained now

[huilo huilo @marcoc2]

@yuzokoshiro better than the one Toe jam and Earl had

[huilo huilo @marcoc2]

@LincolnMargison @_akhaliq this https://t.co/PWk0y9W4bH