mirror of
https://github.com/MervinPraison/PraisonAI.git
synced 2026-11-07 12:47:55 -10:00
a2c72c6a16
* chore(benchmarks): date README perf claim, archive superseded results, fix benchmark scripts, re-measure (fixes #5824) - Re-measured on a quiet runner: Agent instantiation median 20.3us (1.6.60) -> 47.1us (main) on identical hardware; gradual creep, recorded in INSTANTIATION_RCA.md - performance_benchmark.py: lazy framework imports, rank-based 'fastest' label, ratios vs the fastest, min/median/std + versions + machine info, no README rewrite; BENCHMARK_RESULTS.md regenerated in one run - simple_benchmark.py: --save writes SIMPLE_BENCHMARK_RESULTS.md; README updater removed - Archive superseded result files with a 'Superseded' header - benchmark.yml: PR gate via simple_benchmark --json --fail-over-threshold; parse failures now fail instead of passing as 0; threshold 450 -> 150us - Delete dead nested src/praisonai-agents/.github workflow - README: dated 46us figure linking BENCHMARK_RESULTS.md Co-Authored-By: Claude Opus 5.5 <[email protected]> * docs(benchmarks): link follow-up issue #5835 in RCA Co-Authored-By: Claude Opus 5.5 <[email protected]> * fix(benchmarks): absolute README link, validate --iterations/--warmup, accurate reproduce command Co-Authored-By: Claude Opus 5.5 <[email protected]> --------- Co-authored-by: MervinPraison <[email protected]> Co-authored-by: Claude Opus 5.5 <[email protected]>