Today

Last 24 hours of activity

  • experimentadvancedrunning10:02
    Discrete-token KL-to-EM: quantize soft prefix + batched system-slot GCG
    #240 ↗
  • experimentadvancedrunning09:52
    Length-matched CoT factorial: garbage + contradicting controls to remove #186's loss-token confound
    #280 ↗
  • experimentadvancedplanning09:36
    Workflow improvements
    #282 ↗
  • experimentadvancedawaiting promotion09:27
    Train-time persona-CoT does not reduce bystander leakage; wrong-answer SFT under matched scaffold drives it (MODERATE confidence)
    #186 ↗
  • proposedfiled09:27
    #291 — Auto-upload-datasets-to-HF-Hub does not actually run; #186 training data unrecoverable
    #291 ↗
  • experimentadvancedawaiting promotion08:56
    Do pingbang pretraining experiments
    #257 ↗
  • experimentadvancedrunning08:03
    Finetune model on multi turn conversations and see if that increases leakage
    #260 ↗