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
-10
View File
@@ -1409,16 +1409,6 @@ class Gemma4Model(nn.Module, EagleModelMixin):
params_dict.update(dict(self.named_buffers()))
loaded_params: set[str] = set()
for name, loaded_weight in weights:
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[0]
weight_loader(param, loaded_weight)
loaded_params.add(scale_name)
continue
if name.endswith((".k_scale", ".v_scale", ".q_scale", ".prob_scale")):
remapped_name = maybe_remap_kv_scale_name(name, params_dict)
if remapped_name is not None and remapped_name in params_dict: