[Model] Standardize common vision encoders (#31947)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-01-08 02:33:16 -08:00
committed by GitHub
parent d1b6fe007f
commit 5576227bc1
19 changed files with 253 additions and 173 deletions
+2
View File
@@ -404,6 +404,7 @@ class KeyeSiglipAttention(nn.Module):
self.attn = MMEncoderAttention(
num_heads=self.num_heads,
head_size=self.head_dim,
scale=self.scale,
num_kv_heads=self.num_kv_heads,
prefix=f"{prefix}.attn",
multimodal_config=multimodal_config,
@@ -511,6 +512,7 @@ class KeyeSiglipEncoderLayer(nn.Module):
self.mlp = SiglipMLP(
config,
quant_config=quant_config,
multimodal_config=multimodal_config,
prefix=f"{prefix}.mlp",
)