[Model] Use standard ModelOpt config for Inkling NVFP4 (#48990)

Signed-off-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Michael Goin
2026-07-17 18:13:14 -07:00
committed by GitHub
parent fae543015c
commit 02c01f442b
7 changed files with 56 additions and 110 deletions
-1
View File
@@ -73,7 +73,6 @@ class InklingMTPDepthLayer(nn.Module):
is_local=is_local,
quant_config=None,
prefix=f"{prefix}.transformer_block",
nvfp4_config=None,
force_dense_mlp=True,
)