jgrusewski
4911563b2a
feat(D1/N1): temporal self-distillation — snapshot at high health, pull toward best during collapse
- Add q_snapshot.rs: SnapshotRing ring buffer (MAX_SNAPSHOTS=5), health/q_gap admission gate
- Add GpuDqnTrainer::maybe_snapshot_params() — DtoD copy of params_buf into snapshot slot
- Add GpuDqnTrainer::apply_distillation_gradient() — two ungraphed saxpy_f32_aux calls:
grad += alpha * (params - best_snapshot), alpha = 0.1 * (1 - health), skipped when health >= 0.99
- Wire FusedTrainingCtx::maybe_snapshot_qnet() / apply_distillation() / last_distill_active()
- Call from process_epoch_boundary: snapshot when health >= 0.7, distill when health < 0.4
- No new CUDA kernel — reuses existing dqn_saxpy_f32_kernel (saxpy_f32_aux handle)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-20 20:26:36 +02:00
..
2026-04-20 15:29:05 +02:00
2026-04-02 16:29:55 +02:00
2026-04-20 20:26:36 +02:00
2026-04-20 16:59:06 +02:00
2026-04-20 20:04:38 +02:00
2026-03-29 22:35:37 +02:00
2026-03-29 22:35:37 +02:00
2026-04-17 02:05:20 +02:00
2026-03-29 19:18:47 +02:00
2026-04-18 10:19:12 +02:00
2026-04-10 18:54:14 +02:00