Cupra

cupra-ai

10 1 18

https://civitai.com/user/cupra

AI & ML interests

Diffusion models; T2I, T2V, I2V; Naughty things;

Recent Activity

new activity 3 days ago

cicalooo/10Eros-Max-h3-int8-convrot:what's skip edges

liked a model 4 days ago

Qwen/Qwen3.8-27B

liked a model 8 days ago

fal/MiniMax-H3-Realism-People-LoRA

View all activity

Organizations

None yet

New activity in cicalooo/10Eros-Max-h3-int8-convrot 3 days ago

what's skip edges

#1 opened 4 days ago by

thevegit0

liked a model 4 days ago

Qwen/Qwen3.8-27B

Image-Text-to-Text • 28B • Updated 4 days ago • 666k • • 11.2k

liked a model 8 days ago

fal/MiniMax-H3-Realism-People-LoRA

Image-Text-to-Video • Updated 6 days ago • 20.6k • • 255

reacted to onekq's post with 🔥 8 days ago

Post

2657

Looking forward to the Qwen 3.8 model drop, and congratulations on joining the trillion parameters club.

But my eyes are on the promised 27B model. Small (<50B) models decline on OpenRouter, because they are being run on local devices. If people see the family trees here on HF they will understand.

2 replies

reacted to etemiz's post with 🔥 9 days ago

Post

1515

getting ready to fine tune 3.8

- evolutionary strategies
- behavior steering experiments
- expanded dataset
- bringing back ORPO
- more orthogonal evals to keep overfitting minimum
- most probably will take abliterations as base, either mine or somebody else's
- random entropy addition from huggingface fine tunes (take what is popular on hf and randomly introduce into the lineage)
- bring more vibe coding: turns out LLMs know how to fine tune