26 Комити (4e0d85205a62fc52d44a7fd065f422f1329a6561)

Аутор SHA1 Порука Датум
  orangels 70ced01112 feat: record response debug cases in reports пре 1 месец
  orangels a15a62264a fix: set bench max tokens default to 64k пре 1 месец
  orangels 739524628c feat: report effective benchmark request params пре 1 месец
  orangels 986dc4e088 fix: restore bench max tokens default пре 1 месец
  orangels 1a98836b94 fix: set bench max tokens default to 200k пре 1 месец
  orangels 948e32d64a feat: log request errors for debug пре 1 месец
  orangels e0aca08f36 feat: align gpqa scoring with simple evals пре 1 месец
  orangels 898d02f7ef feat: record upstream error debug payloads пре 1 месец
  orangels 18cc7ff716 feat: add provider thinking parameters пре 1 месец
  orangels ee13739c5e feat: add raw response debug mode пре 1 месец
  orangels 60cc009c83 fix: support bare stream flag and cover SSE parsing пре 1 месец
  orangels f21ab8bd4c feat: add --max-tokens CLI param and correct_samples to bench reports пре 1 месец
  orangels 4988a86654 feat: add progress bars to benchmark and RPM execution пре 1 месец
  orangels 527b1e8fc7 feat: add params field to RPM and benchmark reports пре 1 месец
  orangels fdb437a732 fix: correct streaming bugs and RPM diagnose gaps пре 1 месец
  orangels 20217b2250 feat: add streaming (SSE) support with TTFT metrics пре 1 месец
  orangels 2b9fae7ebd feat: report rpm limiter mode details пре 1 месец
  orangels 0a96e11e98 feat: run rpm limiter modes пре 1 месец
  orangels 65ce102004 fix: resolve only selected provider env refs пре 1 месец
  orangels 31c41cb8ac fix: satisfy final clippy checks пре 1 месец
  orangels 83edc789df fix: make benchmark and rpm execution reliable пре 1 месец
  orangels 5d1163d635 feat: run benchmark and rpm tests пре 1 месец
  orangels 1e0926cfba feat: add openai and anthropic request adapters пре 1 месец
  orangels 2064c69924 fix: tighten benchmark dataset helpers пре 1 месец
  orangels 67a4ea8af4 feat: add benchmark dataset fetch and prompts пре 1 месец
  orangels c0ebdea077 chore: initialize cli project skeleton пре 1 месец