← sources

Using compute to verify paper results [D]

source: reddit (reddit:MachineLearning)published 2026-10-29 02:01ingested 2026-10-30 02:37

retained text

I’m not a reviewer yet, but I likely will be by next year.   
As a reviewer, how do you decide to allocate compute to verify paper results? (and is the code in the papers you’re reviewing usually easy to run?)

document fetched 2026-10-30 00:27 · canonical https://reddit.com/r/MachineLearning/comments/1xi7hpf/using_compute_to_verify_paper_results_d