We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 7401cfa commit 75202b0Copy full SHA for 75202b0
src/transformers/models/qwen3_next/configuration_qwen3_next.py
@@ -243,8 +243,10 @@ def __init__(
243
244
self.layer_types = layer_types
245
if self.layer_types is None:
246
+ interval_pattern = kwargs.get("full_attention_interval", 4)
247
self.layer_types = [
- "linear_attention" if bool((i + 1) % 4) else "full_attention" for i in range(self.num_hidden_layers)
248
+ "linear_attention" if bool((i + 1) % interval_pattern) else "full_attention"
249
+ for i in range(self.num_hidden_layers)
250
]
251
layer_type_validation(self.layer_types)
252
0 commit comments