b51eafcfa5
CI / Lint (ruff check) (push) Successful in 28s
CI / Format (ruff format) (push) Successful in 28s
CI / Sync project version with tag (push) Has been skipped
CI / Type check (ty) (push) Successful in 21s
CI / Lint (ruff check) (pull_request) Successful in 25s
CI / Format (ruff format) (pull_request) Successful in 30s
CI / Tests (push) Successful in 1m37s
CI / Sync project version with tag (pull_request) Has been skipped
CI / Type check (ty) (pull_request) Successful in 31s
CI / Tests (pull_request) Successful in 55s
Trains the routed trunk's forward combination as a hard one-hot sample (matching eval-time top-1 dispatch exactly) while keeping a smooth gradient on the backward pass, targeting the train/eval mismatch identified as a likely contributor to experts overlapping instead of partitioning in the first energy-router rollout benchmark. Off by default (model.router.gumbel); existing routed configs/checkpoints are unaffected. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>