Alex Open Research Wiki

Tag: zeroth-order

9 items with this tag.

  • Oct 25, 2026

    Dust

    • activation-perturbation
    • zeroth-order
    • pretraining
  • Oct 25, 2026

    Noise

    • benchmark
    • speedrun
    • zeroth-order
  • Oct 25, 2026

    SOMA

    • recurrent-models
    • ensembles
    • zeroth-order
  • Oct 25, 2026

    1.5-SPSA: Probe-Space Preconditioning for Fast and Stable Zero-Order Training

    • zeroth-order
    • optimization
    • preconditioning
    • post-training
    • benchmark-hygiene
  • Oct 25, 2026

    Dust: Pretraining Transformers Without Backpropagation

    • zeroth-order
    • activation-perturbation
    • credit-assignment
    • pretraining
  • Oct 25, 2026

    Noise: Speedrun for Pretraining without Backpropagation

    • zeroth-order
    • benchmark
    • speedrun
    • pretraining
    • compute-accounting
  • Oct 25, 2026

    SOMA: Scaling Zero-Order Pretraining through Model Sharding

    • zeroth-order
    • recurrent-models
    • ensembles
    • credit-assignment
    • training-efficiency
  • Oct 25, 2026

    Scaling Recurrent Neural Networks to a Billion Parameters with Zero-Order Optimization

    • zeroth-order
    • recurrent-models
    • bptt
    • credit-assignment
  • Oct 25, 2026

    Forward-Only Sequence Training

    • zeroth-order
    • credit-assignment
    • recurrent-models
    • bptt
    • training-efficiency

Created with Quartz v4.5.2 © 2026