foxhunt

Files

jgrusewski 0e70bf96fe fix(rl): wire BCE + aux losses into GPU loader path + event-based sync

BCE and aux losses were hardcoded to 0 in the GPU data path. Now
step_batched_from_device returns (bce_loss, aux_loss), stored as
last_bce_loss / last_aux_loss and fed into step stats + JSONL.

Event-based sync replaces host-blocking stream sync:
- perception: raw_event_record after training graph, raw_event_sync
  in read_deferred_stats (instant — graph completed long ago)
- diag_staging: raw_event_record after snapshot_async, raw_event_sync
  in sync_and_swap (instant — copies completed during GPU work)

All JSONL fields preserved. Loss values one-step deferred (same as
before for Q/π/V, now also for BCE/aux).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

2026-05-26 22:18:30 +02:00

cuda

fix(rl): wire BCE + aux losses into GPU loader path + event-based sync

2026-05-26 22:18:30 +02:00

examples

fix(rl): wire BCE + aux losses into GPU loader path + event-based sync

2026-05-26 22:18:30 +02:00

src

fix(rl): wire BCE + aux losses into GPU loader path + event-based sync

2026-05-26 22:18:30 +02:00

tests

perf(rl): fuse adamw_increment into adamw_step + tau into iqn_forward