Skip to content

Add Baseline -> GARI -> Multi-Pass tradeoff benchmarks and plots - #323

Open
aria-googler wants to merge 1 commit into
mainfrom
multipass-tradeoff-plots
Open

aria-googler wants to merge 1 commit into
mainfrom
multipass-tradeoff-plots

Conversation

@aria-googler

Copy link
Copy Markdown
Contributor

Summary

Adds end-to-end tradeoff benchmarks and log–log plots (Logical Error Rate per Round vs. Decode Time per Round) showing the progression across the three correlated decoding stages in Tesseract:

  • Baseline (Monolithic Tesseract): beam=20, beam_climbing=True, no_revisit_dets=True, num_det_orders=1
  • GARI: prior=xor, beam=5, beam_climbing=True, num_det_orders=1
  • Multi-Pass: 2-pass causal scheduling (--multipass --num-passes 2 --multipass-strategy causal), evaluated at beam=20 and beam=5 (beam_climbing=True, no_revisit_dets=True, num_det_orders=1)

Contents

  • README.md: Embeds the combined progression plot in the Multi-Pass Graph Shattering section and links to benchmarking/multipass/README.md.
  • benchmarking/multipass/README.md: Full breakdown of the combined, Superdense Color Code (d=3, 5, 7, 9), and Bivariate Bicycle ([[144, 12, 6]]) tradeoff plots, summary table, and reproduction instructions.
  • benchmarking/multipass/run_benchmarks.py & benchmarking/multipass/make_plots.py: Unified benchmark runner and plot generator.
  • benchmarking/multipass/results.json & benchmarking/multipass/plots/*.png: Benchmark results and generated plots.

@aria-googler
aria-googler requested a review from a team as a code owner September 25, 2026 22:30
@aria-googler
aria-googler requested review from oscarhiggott and removed request for a team September 25, 2026 22:30

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant