4 ms·
Codex reasoning-token clustering at 516 may be leading to degraded performance
- cyanydeez 3mo agowonder if theyre basically doing this llamacpp reasoning trick: https://github.com/ggml-org/llama.cpp/blob/master/common/reasoning-budget.cpp https://github.com/ggml-org/llama.cpp/blob/master/common/rea... id guess its the harness
- virengupta45 3mo agothis is happening on xhigh config!! extremely pissing off