Pinned post
Last week, I said every parser is #1 on its own benchmark. So we tested the benchmarks.
We ran 8 parsers across 12 benchmarks, then tilted the pages by 2–4°. 6 benchmarks got a different winner.
We’re calling it BenchBench.
We ran 8 parsers across 12 benchmarks, then tilted the pages by 2–4°. 6 benchmarks got a different winner.
We’re calling it BenchBench.