Dan M

Work Hard. Play Harder!

[Dan M @SDGameGuy]

@jarredsumner @curlykoder @zack_overflow @djgrant_ @jarredsumner What was the total API cost? I think the Claude generated C compiler was around $20k? If you don't know right now, please include in the blog post!

[Dan M @SDGameGuy]

@OwenDugan @OwenDugan "The encoder–decoder design gets us to the information-theoretic limit" - Do you mean beyond the asymptotically? The graph looks like it requires ~10-25x the number of parameters per fact?

[Dan M @SDGameGuy]

@cloneofsimo @threejs What was your prompt?

[Dan M @SDGameGuy]

@pr0cf51 @TeamAtlanta24 Any timeframe for releasing the custom model?

[Dan M @SDGameGuy]

@augmentcode The best dev plugin is getting a multiplier.

[Augment Code @augmentcode]

Remote Agent is live in VS Code (v0.472.1).

Early users are already shipping while they sleep. One dev launched 3 agents before logging off - woke up to a PR, docs, and a fixed flaky test.

We’re celebrating wins like that by giving 5 devs 4 months of Augment Max ($1K in value each).

RT + reply with your worst backlog task. Every 100 RTs = one more winner

[Dan M @SDGameGuy]

@GosuCoder @hebbarmp Please let us know how it goes.

[Dan M @SDGameGuy]

@jrysana So...
{
"type": "object",
"properties": {
"query": { "type": "string", "description": "User question" },
"answer": { "type": "string", "description": "System answer" }
},
"required": ["query", "answer"]
}

:-P

[Dan M @SDGameGuy]

@jrysana Still cool, but damn... was hoping you cracked AI too cheap to meter :-) Are your prior examples what we can expect for 1 billion RPM, or do you have a newer one?

[Dan M @SDGameGuy]

@attentionmech You could even call it a superposition ;-)

[Dan M @SDGameGuy]

@jrysana If it was per user, you'd probably become the #1 cause for global warming :-P

[Dan M @SDGameGuy]

@jrysana "general AI" to me means ~ OpenAI/Claude/Grok equivalent. Is that what your claiming at 1 billion RPM (soon)?

[Dan M @SDGameGuy]

@jrysana It seems it's still a waitlisted? I mean public GA. Don't think I ever got a response (even confirmation) after joining waitlist...

[Dan M @SDGameGuy]

@natolambert I haven't seen this with o3-high or gemini 2.5 pro. The longer the conversation, they worse they are at instruction following, make more mistakes. What strongest models are you referring to?

[Dan M @SDGameGuy]

@_arohan_ Is the source for per layer embeddings available? I can't find the source for per layer embeddings in mediapipe or litert on github, and the 3n hf model is just a .task , so no hf transfomers...

[Dan M @SDGameGuy]

@attentionmech Request: include a link to papers you reference. Maybe in self-reply if you're worried a link will reduce your reach. Or broken link if you're extra paranoid:
arxiv. org/abs/2410.01545

@rohanpaul_ai (who often posts about interesting AI papers) does this

[Dan M @SDGameGuy]

@XorDev Keep this up and you're going to accidently solve quantum computing

[Dan M @SDGameGuy]

@__tensorcore__ CuTeDSL EULA seems to be not open source, or compatible with GPL. Given how tied it is to is to Nvidia hardware and runtime, it seems like an unncessary restriction...?