log attn mask

DavidLandup0 · DavidLandup0 · commit c45f589d2fd9 · 2025-07-26T16:26:33.000+09:00
diff --git a/keras_hub/src/models/smollm3/smollm3_utils.py b/keras_hub/src/models/smollm3/smollm3_utils.py
@@ -47,6 +47,7 @@ def eager_attention_forward(
     )
 
     # Apply attention mask if provided
+    print("attention_mask", attention_mask)
     if attention_mask is not None:
         causal_mask = attention_mask[:, :, :, : ops.shape(key_states)[-2]]
         attn_weights = ops.add(attn_weights, causal_mask)

Original file line number	Diff line number	Diff line change
`@@ -47,6 +47,7 @@ def eager_attention_forward(`
`47`	`47`	`)`
`48`	`48`
`49`	`49`	`# Apply attention mask if provided`
	`50`	`+ print("attention_mask", attention_mask)`
`50`	`51`	`if attention_mask is not None:`
`51`	`52`	`causal_mask = attention_mask[:, :, :, : ops.shape(key_states)[-2]]`
`52`	`53`	`attn_weights = ops.add(attn_weights, causal_mask)`