blurb

Understand a paper or concept

Paste an arxiv link or name an idea, then work through an interactive explainer.

Enter to generate, Shift + Enter for a new line

Recent explainers

11

  1. How are AI agents used? Evidence from 177,000 MCP tools — monitoring the action space

    2603.2380218 min read

  2. GeoSANE — generating geospatial models from weights

    2603.2340815 min read

  3. Continual Learning Bench — isolating improvement from capability

    2606.0566116 min read

  4. End-to-End Test-Time Training for Long Context

    2512.2367516 min read

  5. Continual Learning Bench — isolating learning from capability

    2606.0566118 min read

  6. Layer normalization — stabilizing deep activations across batch sizes

    concept16 min read

  7. ReOPD — offline multi-turn distillation without environment rollouts

    2607.0476318 min read

  8. Batch normalization — stabilizing internal covariate shift

    concept15 min read

  9. cross-entropy loss — from information theory to gradient flow

    concept17 min read

  10. KL divergence — why the direction you minimize changes what you get

    concept16 min read

  11. Attention Is All You Need — how deleting recurrence still won

    1706.0376218 min read