ankit219
- Karma
- 0
- Created
- ()
- Submissions
- 0
- Mythos/Fable-5 is a greedy Depth First Search system (ankitmaloo.com)
- How Anthropic trained Fable 5 => by analysing its reasoning traces (ankitmaloo.com)
- LLMs give us a way to factorize intelligence (ankitmaloo.com)
- Show HN: Open-Source SDK for AI Knowledge Work (github.com)
GitHub: https://github.com/ClioAI/kw-sdk Most AI agent frameworks target code. Write code, run tests, fix errors, repeat. That works because code has a natural verification signal. It works or it doesn't. This SDK…
- World Models (ankitmaloo.com)
- Every big lab is putting resources in building world models (ankitmaloo.com)
- Using coding LLM agents to hack Catan's browser game (ankitmaloo.com)
- RL environments need to account for changing priors to work (ankitmaloo.com)
- Notes on RL Environments (ankitmaloo.com)
- Show HN: I created a random vibe coding cost estimation site via lovable (preview--ai-lever-finance.lovable.app)
- Bitter Lesson is about AI agents (ankitmaloo.com)
- Apple's motion to dismiss their antitrust lawsuit [pdf] (storage.courtlistener.com)
- Apple allowing 3rd party payments in Netherlands w commission(2022) (daringfireball.net)
- Luttig on narratives currently prevalent in AI discourse (luttig.substack.com)