Under double-blind review

Download code and data How it was checked

Section 4

How often, and by which bar

Bar
Section 5

When the input says less

Section 6

After it is written

Section 7

A write-time fix

Section 8

Does collapsed memory lead to wrong actions?

Figures

Every figure in the paper

Tables

Every table in the paper

Code and data

Everything needed to check the numbers

    pip install numpy scipy matplotlib pillow python-dotenv openai
python3 scripts/make_provenance_paper.py
python3 scripts/fig1_arch.py provenance
python3 scripts/fig_ladders.py provenance
python3 scripts/fig_prov_story.py
python3 scripts/fig_prov_3d.py
python3 scripts/check_provenance_facts.py
    Limitations

    From the paper