v2; a lot of hacks

2025-10-15 18:02:10 +09:00
parent e51017897d
commit 6ab33ceb83
7 changed files with 586 additions and 199 deletions
--- a/src/model/dino.py
+++ b/src/model/dino.py
@@ -235,7 +235,7 @@ class DINOv3ViTAttention(nn.Module):
            key_states,
            value_states,
            dropout_p=dropout_p,
-            softmax_scale=None,  # Will use default 1/sqrt(headdim)
+            softmax_scale=None,
            causal=False,
        )