Readme
#1
by
chemuduguntar
- opened
Hi there,
Can you tell us a bit about what this model does? Does it generate openscad code given images and video?
TLDR: Dont use it. Current sota models are way stronger than this even in the niche.
It was a failed research attempt at SFT finetuning qwen to generating openscad code from, images and or text.
It mostly learned the openscad syntax, but actual 3d modeling was way off.
Will maybe try this again, with added reinforcement learning on qwen3vl, but scrapped for now.
Thanks for replying back, you are right the SOTA models are getting quite good at this, and don't need any SFT right now.
chemuduguntar
changed discussion status to
closed