continuous_cvae-Apr26-reconstruction-stack-followup_20260426

Status: results drafted; repo-side aggregate and curated observational results note are complete. Interpretive analysis and cross-collection deep dives are still pending.

This collection is a targeted follow-up to continuous_cvae-Apr24-reconstruction-stack_20260424. It keeps the Apr24 task, adaptation-mode, model, and training grid, but narrows the reconstruction menu to the unresolved cells needed to interpret the Apr24 frontier: cc-TWR-only variants, example-inverse-surprisal plus cc-TWR, low-beta raw controls, and a beta-0.1 continuation of the strongest distribution-normalized plus cc-TWR setting.

Artifacts

  • Family spec: path:experiments/synthetic_sequences/slurm/families/disentangled_from_entangled_random_init/specs/continuous_cvae_apr26_reconstruction_stack_followup.yaml
  • Shared reconstruction menu: path:experiments/synthetic_sequences/slurm/families/shared/reconstruction_stack_followup_sweep.py
  • Slurmkit collection: path:.slurmkit/collections/synthetic_sequences/families/disentangled_from_entangled_random_init/continuous_cvae_apr26_reconstruction_stack_followup.yaml
  • Raw run directory: path:.runs/synthetic_sequences/disentangled/exp2_continuous_cvae-Apr26-reconstruction-stack-followup
  • Analysis brief: path:experiments/synthetic_sequences/disentangled/analysis/aggregates/continuous_cvae-Apr26-reconstruction-stack-followup_20260426/brief.yaml
  • Aggregate output directory: path:experiments/synthetic_sequences/disentangled/analysis/aggregates/continuous_cvae-Apr26-reconstruction-stack-followup_20260426
  • Repo-side report: path:experiments/synthetic_sequences/disentangled/analysis/aggregates/continuous_cvae-Apr26-reconstruction-stack-followup_20260426/report.md
  • Machine-readable findings: path:experiments/synthetic_sequences/disentangled/analysis/aggregates/continuous_cvae-Apr26-reconstruction-stack-followup_20260426/findings.json

Context

Runs

  • Slurmkit collection name: synthetic_sequences/families/disentangled_from_entangled_random_init/continuous_cvae_apr26_reconstruction_stack_followup
  • Training collection name: exp2_continuous_cvae-Apr26-reconstruction-stack-followup
  • Planned grid: 6 tasks x 9 loss settings x 3 adaptation modes x 1 seed = 162 jobs
  • Completed grid: 162 / 162 jobs in the current aggregate.
  • Analysis stance: use the slurmkit job grid as the expected denominator and keep the aggregate script idempotent so figures can be regenerated if artifacts change.
Built with LogoFlowershow