gradient checkpointing

2026-06-04 14:09:14 +02:00
parent f2b501d8c9
commit 45269d16bb
4 changed files with 23 additions and 6 deletions
@@ -38,6 +38,7 @@ meta:
  pred_emb_dim: 384
  read_checkpoint: null
  use_bfloat16: true
+  use_gradient_checkpointing: true
 optimization:
  ema:
    - 0.996