Deschide articolul original într-un tab nou Schimbarea acestei opțiuni se aplică imediat și reîncarcă pagina.
LECTURA
← Înapoi la Urmăritori

Alex Wa's Blog

djdumpling.github.io/feed.xml
Vizitează site-ul

Cel mai recent articol

Ultima verificare

Urmăritori 0 urmăritori
Articole: 6

  1. Learning MegaGem, from self-play to price discovery

    (Alex Wa's Blog)

    We study MegaGem, a three-player, general-sum, imperfect-information auction game developed by Jane Street, and train a 4B specialist to play it.

  2. hunting for headroom on modded-nanoGPT (WR #82)

    (Alex Wa's Blog)

    Modded-nanoGPT speedrunning is a popular benchmark that asks how fast 8xH100s can drive a 124M-parameter model to 3.28 validation loss on FineWeb.

  3. frontier model training methodologies

    (Alex Wa's Blog)

    How do labs train a frontier, multi-billion parameter model?

  4. activation engineering for privacy protection in LLMs

    (Alex Wa's Blog)

    LLMs trained on web-scale corpora inadvertently memorize and leak personally identifiable information (PII) present in their training data.

  5. combinatorial reasoning environments for LLMs and RL

    (Alex Wa's Blog)

    Can RL agents learn to play spatial reasoning puzzle games as well as, or better than, LLMs?

  6. whirlwind of PPO and RLHF for LLMs from scratch

    (Alex Wa's Blog)

    RLHF with PPO from scratch and lots of fine-tuning GPT-2 models for movie sentiment classification.