mrq/torchscale
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

fix bug

Browse Source
This commit is contained in:
sunyt32 2023-09-28 17:05:53 +00:00
parent fd8234c2ac
commit 59fc5f7d3d
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
torchscale/component/multiscale_retention.py
View File

@ -63,7 +63,7 @@ class MultiScaleRetention(nn.Module):
self.out_proj = MultiwayWrapper(args, nn.Linear(value_dim, embed_dim, bias=False))
self.group_norm = MultiwayWrapper(args, RMSNorm(self.head_dim, eps=args.norm_eps, elementwise_affine=False))
self.group_norm = MultiwayWrapper(args, RMSNorm(self.head_dim, eps=args.layernorm_eps, elementwise_affine=False))
self.reset_parameters()
def reset_parameters(self):