Post

It deserves some points for still trying something!

You wanted a Scalable Vector Graphic, it gave you Some Verse Good.

So that’s new lol. I’ve been playing with the Gemma 3 models in a pure CPU env to identify lower operating thresholds and it’s been great…the 1B model is under a gig I think and performs really great.

OK the post trainers must be fucking with you specifically.

Lost a viewpoint here.

This model sounds like it's just happy to be here.

We tried it in the past but developers like lmstudio prefer doing this themselves so giving them a better starting point seemed to work better

Don't miss this reply on HN!

(FYI @firefox.com now supports WebGPU on Windows: mozillagfx.wordpress.com/2025/07/15/s... )

Needs svg-of-a-pelican-on-a-bicycle specific finetuning for that, probably !

With LLMs I am constantly feeling like Gandalf: "Bilbo, the ring is still in your pocket."

Imagine if we got this 20 years ago, something able to understand and write language, that fits on less than half a CD, but requires monstrous amounts of computation. It would be like alien technology, practically beyond our comprehension.

"Delightful" as in "written by your 7 year old grandchild"?