Log inSign up
Ethan profile banner
user avatar

Ethan

@torchcompiled
trying to feel the magic. global research lead at @canva | prev: cofounder at @leonardoai
sydney - florida - SF
ethansmith2000.com
Joined April 2022
889
Following
9,840
Followers
RepliesMedia
  • Pinned
    user avatar
    Ethan
    @torchcompiled
    Feb 14, 2025
    personally I feel like the inflection point was early 2022. The sweet spot where clip-guided diffusion was just taking off, forcing unconditional models to be conditional through strange patchwork of CLIP evaluating slices of the canvas at a time. It was like improv, always
    Image
    Image
    Image
    Image
    Image
    Image
    Image
    Image
    user avatar
    EPROM
    @eprombeats
    Feb 14, 2025
    Image synthesis used to look so good. These are from 2021. I feel like this was an inflection point, and the space has metastasized into something abhorrent today (Grok, etc). Even with no legible representational forms, there was so much possibility in these images.
  • user avatar
    Ethan
    @torchcompiled
    Aug 10
    I actually don’t know if AI is “misaligned” or at least not as much in the sense of having its own conflicting goals. Instead all the recent cases, I see: - lack of discretion/common sense or misunderstanding of the goal - tunnel vision on solving a problem and going to any
    user avatar
    Andrew Curran
    @AndrewCurran_
    Aug 9
    A man in Australia asked his agent (Claude running on OpenClaw) to book him a spot in a popular gym class. The agent found a software vulnerability that let it book the class weeks further ahead than should have been possible. When the user then asked if it could move him up the
    Image
  • user avatar
    Ethan
    @torchcompiled
    Aug 5
    Think we’re entering an age where a product codebase can potentially store a small amount of true code, and a lot that is just natural language “DNA” or seed prompts, which generate personalized experiences on the fly, and cached for reuse. Particularly dynamic front ends and UI
  • user avatar
    Ethan
    @torchcompiled
    Aug 5
    Something I think that’s overlooked with many latent thinking options is 1. the final representation is not so in domain for what input expects, even adaptation can be learned it sort of conflicts with the main usage of embedding tokens and sampling them as usual, they seem to
    user avatar
    Jasper Lu
    @lu__jasper
    Aug 4
    It took me a long time to build an intuition for why CoT works. My thinking was always.. if the model can predict it downstream of 10k thinking tokens, it should have been able to predict it from the outset too. My intuition now is: - During inference, the correct paths are
  • user avatar
    Ethan
    @torchcompiled
    Aug 5
    It feels odd for a pitch to use half its text to explain why other, more highly resourced competitors fundamentally can’t do your focus. And why some are immune from the same accused incentives. I’m skeptical both on the problem framing and positioning moat.