[Bugfix] Exclude vision embedder from quantization in Gemma4 Unified (#44571)
Signed-off-by: Luciano Martins <lucianommartins@users.noreply.github.com> Co-authored-by: Luciano Martins <lucianommartins@users.noreply.github.com>
This commit is contained in:
co-authored by
Luciano Martins
parent
4efd6ffde0
commit
da1daf40bf
@@ -80,7 +80,7 @@ class Gemma4UnifiedVisionEmbedder(nn.Module):
|
|||||||
Pipeline: raw patches → LN₁ → Dense → LN₂ → +factorized_posemb → LN₃.
|
Pipeline: raw patches → LN₁ → Dense → LN₂ → +factorized_posemb → LN₃.
|
||||||
"""
|
"""
|
||||||
|
|
||||||
def __init__(self, config, quant_config=None):
|
def __init__(self, config, quant_config=None, prefix=""):
|
||||||
super().__init__()
|
super().__init__()
|
||||||
patch_dim = config.model_patch_size**2 * 3
|
patch_dim = config.model_patch_size**2 * 3
|
||||||
mm_embed_dim = config.mm_embed_dim
|
mm_embed_dim = config.mm_embed_dim
|
||||||
@@ -91,6 +91,7 @@ class Gemma4UnifiedVisionEmbedder(nn.Module):
|
|||||||
mm_embed_dim,
|
mm_embed_dim,
|
||||||
bias=True,
|
bias=True,
|
||||||
quant_config=quant_config,
|
quant_config=quant_config,
|
||||||
|
prefix=f"{prefix}.patch_dense",
|
||||||
gather_output=True,
|
gather_output=True,
|
||||||
)
|
)
|
||||||
self.patch_ln2 = nn.LayerNorm(mm_embed_dim)
|
self.patch_ln2 = nn.LayerNorm(mm_embed_dim)
|
||||||
@@ -267,6 +268,7 @@ class Gemma4UnifiedForConditionalGeneration(Gemma4ForConditionalGeneration):
|
|||||||
Gemma4UnifiedVisionEmbedder(
|
Gemma4UnifiedVisionEmbedder(
|
||||||
config.vision_config,
|
config.vision_config,
|
||||||
quant_config=quant_config,
|
quant_config=quant_config,
|
||||||
|
prefix=maybe_prefix(prefix, "vision_embedder"),
|
||||||
)
|
)
|
||||||
if config.vision_config is not None
|
if config.vision_config is not None
|
||||||
else None
|
else None
|
||||||
|
|||||||
Reference in New Issue
Block a user