Chromadera retweeted
Ref version of the Frieren LoRA for MiniMax H3 is out!
Trained on the ref model (with @ostrisai’s ref training adapter v3, and musubi fork), so it works better with ref2va than the FL2VA One. Same character sheets + voice refs from the HF repo below.
Let me know what you think!
huggingface.co/tazmannner/Fr…
released my Frieren LoRA for MiniMax H3
every scene in the video is H3 + the LoRA. colors, line work, animation timing and even the bgm are from the show's style now
also included: 37 character sheets, generated voice refs for 27 characters, a prompt guide and the full training write-up. My most ambitious style lora yet:
huggingface.co/tazmannner/Fr…
Chromadera retweeted
Replying to @tazmannner @ostrisai
I recommend the 3600 steps checkpoint, after this validation started to plateau and trend negatively. This is trained on the same exact parameters, including audio as the FLV2A lora. I didn't extensively test it though, and I just generated this samples and called it a day. If I find any issues or a different checkpoint is better I'll share. I wanted to get this out the door quickly so I could work on other things. Hope it's helpful!
Chromadera retweeted
Gemini 4 Carbon - Rendering a beautiful woman. Doesn't she look like that spy who had an affair with Jensen?
making your own software is tough, even with AI, because it's hard to know what you really want
I didn't end up sticking with this, the modes weren't the right way for me to organize my work
gonna a different take that’s more to-do oriented
Chromadera retweeted
big video, small model.
Rumble-2-Flash is now open weights.
40+ languages, Apache 2.0, 0.5B parameters.
download it: huggingface.co/Reson8/Rumble…
Chromadera retweeted
Replying to @Jalann_art
Don't get me started on how everyone who's Black or White HAVE to be represented as American. No other ethnicity matters.
Chromadera retweeted
Replying to @itsmissglitch
They don't have to make the best coding model either. I'll be happy with a slightly smarter Gemma 31b but much better tool calling. Something that can drive my custom telegram harness. It's so much better to talk to than current models, frontier or otherwise
Chromadera retweeted
说个暴论,真搞不懂现在怎么还有那么多教人用 AI 的教程。
现在的 AI 早就不是前几年那种理解力有限,非得背一套提示词模板才能干活的时候了。打开软件,直接用日常语言像聊天一样把想法说清楚不就行了吗?连正常说话都要看教程学,缺的根本不是什么工具技巧,就是基本的表达和沟通能力。
真正该花时间去学的,反而是专业本身的底子。就拿写代码来说,现在的门槛根本不在怎么用 AI,而在软件工程知识。项目结构怎么拆合理,边界条件怎么兜底,还有 AI 生成的代码到底靠不靠谱、有没有暗坑,这些全得靠自己的工程能力去判断。放着正经的软件工程不学,天天研究怎么在聊天框里提问,纯属本末倒置。
Chromadera retweeted
Well chaps, that was certainly more fun than a three-day symposium on emerging market equities.
Chromadera retweeted
my multiplayer arcade volleyball game 'just volleyball' is OUT NOW🏐
catch the launch discount and pick it up for $4.95 while you can!
Chromadera retweeted
In a year from now people are gonna be saying you're NGMI if still using markdowns for your agents btw
Chromadera retweeted
This light is driving me crazy.
I gave Gemini 4 Carbon a shot through @factory and honestly, it’s a good model. This scene came out of 1 prompt with zero fixes.
Every candle is a real light. Drag one and the shadows swing, the wine throws red caustics across the cloth, and the glints on the glass are insane. 1 HTML file, no assets, all of it built in code.
YuE2 music on my RTX 3090






























