@@ -43,9 +43,13 @@ def resolve_hf_cache_layout(config):
4343
4444 if layer_types is None :
4545 if getattr (text_config , "sliding_window" , None ) is not None :
46- layer_types = ["sliding_attention" for _ in range (text_config .num_hidden_layers )]
46+ layer_types = [
47+ "sliding_attention" for _ in range (text_config .num_hidden_layers )
48+ ]
4749 else :
48- layer_types = ["full_attention" for _ in range (text_config .num_hidden_layers )]
50+ layer_types = [
51+ "full_attention" for _ in range (text_config .num_hidden_layers )
52+ ]
4953 else :
5054 layer_types = list (layer_types )
5155
@@ -54,13 +58,20 @@ def resolve_hf_cache_layout(config):
5458
5559 if hasattr (text_config , "global_head_dim" ):
5660 head_dims = [
57- text_config .global_head_dim if layer_type == "full_attention" else text_config .head_dim
61+ (
62+ text_config .global_head_dim
63+ if layer_type == "full_attention"
64+ else text_config .head_dim
65+ )
5866 for layer_type in layer_types
5967 ]
6068 num_heads = [
61- text_config .num_global_key_value_heads
62- if layer_type == "full_attention" and getattr (text_config , "attention_k_eq_v" , False )
63- else text_config .num_key_value_heads
69+ (
70+ text_config .num_global_key_value_heads
71+ if layer_type == "full_attention"
72+ and getattr (text_config , "attention_k_eq_v" , False )
73+ else text_config .num_key_value_heads
74+ )
6475 for layer_type in layer_types
6576 ]
6677 else :
0 commit comments