Ostris's banner

Ostris

@ostrisai • 13,010 subscribers

AI / ML researcher and developer. Creator of AI Toolkit - https://t.co/Thqof0Gxpj Support my work - https://t.co/Isg2EXrP7s

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

I trained this LTX LTX 2.3 LoRA of George Costanza at home on my 5090 in about a day with AI Toolkit. I generated this 30 second video with ComfyUI on my 5090 in 6 minutes. Open source is, always has been, and always will be, the future of generative AI. (SOUND ON)

I trained this LTX LTX 2.3 LoRA of George Costanza at home on my 5090 in about a day with AI Toolkit. I generated this 30 second video with ComfyUI on my 5090 in 6 minutes. Open source is, always has been, and always will be, the future of generative AI. (SOUND ON)

390,565 views • 2 months ago

I trained an ACEStep 1.5 XL LoRA on "some obscure 60s English rock band". Then I wrote a song about LoRA training and had them play it. Absolutely wonderful experience. I still have some UI work before I can make training public in AI Toolkit, but working on it as fast as I can.

I trained an ACEStep 1.5 XL LoRA on "some obscure 60s English rock band". Then I wrote a song about LoRA training and had them play it. Absolutely wonderful experience. I still have some UI work before I can make training public in AI Toolkit, but working on it as fast as I can.

36,597 views • 2 months ago

I added the ability to merge LoRAs in AI Toolkit. I always find a merge of the last few checkpoints yields better results. Basically a "poor man's EMA". But more importantly, this change adds the framework to run small scripts from the frontend, which I have a lot of plans for.

I added the ability to merge LoRAs in AI Toolkit. I always find a merge of the last few checkpoints yields better results. Basically a "poor man's EMA". But more importantly, this change adds the framework to run small scripts from the frontend, which I have a lot of plans for.

11,492 views • 22 days ago

I trained another one on some obscure 90s English rock band. Same prompt and song, different performers. Someone needs to build desktop software that generates lyrics by shuffling customizable prompts, shuffles and mixes custom LoRAs, and queues up songs.

I trained another one on some obscure 90s English rock band. Same prompt and song, different performers. Someone needs to build desktop software that generates lyrics by shuffling customizable prompts, shuffles and mixes custom LoRAs, and queues up songs.

10,452 views • 2 months ago

AI Toolkit now officially supports training Lightricks LTX-2 LoRAs.

AI Toolkit now officially supports training Lightricks LTX-2 LoRAs.

18,652 views • 4 months ago

Testing Flux.1 schnell embedding interpolation. I told claude some details about my life and had it generate 40 prompts from being born -> being dead using the data I gave it. Then interpolated between the embeddings for this video. Uses Araminta softpasty LoRA

Testing Flux.1 schnell embedding interpolation. I told claude some details about my life and had it generate 40 prompts from being born -> being dead using the data I gave it. Then interpolated between the embeddings for this video. Uses Araminta softpasty LoRA

45,740 views • 1 year ago

LTX-2 by Lightricks is amazing.

LTX-2 by Lightricks is amazing.

12,085 views • 5 months ago

Qwen-Image pruning experiment. Going from 60 to 30 blocks, 20B params to 10B params. Removed block idx 2, 3, 4, 5, 7, 8, 10, 11, 12, 13, 14, 15, 16, 21, 23, 24, 40, 41, 42, 43, 44, 45, 49, 50, 51, 52, 53, 54, 55, 56

Qwen-Image pruning experiment. Going from 60 to 30 blocks, 20B params to 10B params. Removed block idx 2, 3, 4, 5, 7, 8, 10, 11, 12, 13, 14, 15, 16, 21, 23, 24, 40, 41, 42, 43, 44, 45, 49, 50, 51, 52, 53, 54, 55, 56

14,118 views • 9 months ago

Wan 2.1 14B LoRA training working on 24GB! 🥳 Avg 1.7sec/step at 480p on a 4090. LR was too low on this one.

Wan 2.1 14B LoRA training working on 24GB! 🥳 Avg 1.7sec/step at 480p on a 4090. LR was too low on this one.

16,100 views • 1 year ago

No more content to load