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.
attn_temperature_tuning
1 parent 0d06b53 commit 1afa994Copy full SHA for 1afa994
vllm/model_executor/models/llama4.py
@@ -148,9 +148,8 @@ def __init__(self,
148
self.q_size = self.num_heads * self.head_dim
149
self.kv_size = self.num_kv_heads * self.head_dim
150
self.scaling = self.head_dim**-0.5
151
- # TODO: attn_temperature_tuning should be a bool in huggingface
152
self.attn_temperature_tuning = self.nope and \
153
- config.attn_temperature_tuning > 0
+ config.attn_temperature_tuning
154
155
self.floor_scale = getattr(config, "floor_scale", 8192.0)
156
self.attn_scale = getattr(config, "attn_scale", 0.1)
0 commit comments