Documentation
¶
Overview ¶
Command loopsweep finds the smallest loop-budget work (gpubatch layer 4) with which a set of e2e tests still passes with zero truncations: it runs them at LoopWork = 2^k for k from --from to --to, with the per-loop floor at 1 so that the floor masks nothing, and prints one line per k. The smallest passing k, against the shipped default (gpubatch.DefaultLoopWork), is the margin of the default.
loopsweep --from 24 --to 38 --run 'TestLLM(Oracle|Trace|Shaders)' \
--env LLM_MODELS=... --env LLM_REPO=... --out DIR
Click to show internal directories.
Click to hide internal directories.