Ofir Press @OfirPress · 27 Aug 2026
Yes- the gold mine that @closji & @jyangballin discovered in SWE-bench is that if you find a scalable way to mine benchmark tasks, not only will you have an easier time building the benchmark, but you'll also have a straightforward method to getting lots of training examples!
3 074Views
13Likes
2Reposts
3Replies
0Quotes
5Bookmarks
Is that a lot?
0.96×vs this author's median3 207 views is typical
27Percentile for this authorof 11 recent posts
0.85×vs 10K–100K median3 598 views is typical
15.77%Reachviews ÷ followers
0.59%Engagement rateof viewers reacted