Remove KV cache scale boilerplate from model weight loading methods (#43167)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2026-06-05 05:19:04 -07:00
committed by GitHub
co-authored by mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
parent 7fe7800fa4
commit 62215e72c6
56 changed files with 87 additions and 730 deletions
-15
View File
@@ -790,21 +790,6 @@ class KeyeSiglipVisionModel(nn.Module):
continue
if "head.mlp" in name or "head.probe" in name:
continue
if self.quant_config is not None and (
scale_name := self.quant_config.get_cache_scale(name)
):
param = params_dict[scale_name]
weight_loader = getattr(
param,
"weight_loader",
default_weight_loader,
)
loaded_weight = (
loaded_weight if loaded_weight.dim() == 0 else loaded_weight[0]
)
weight_loader(param, loaded_weight)
loaded_params.add(scale_name)
continue
for (
param_name,
weight_name,