riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-31 02:02:21 +06:00
Code Issues Actions 23 Packages Projects Releases Wiki Activity

Fix Switch Transformers When sparse_step = 1 (#28564)

Browse Source 
Fix sparse_step = 1

I case sparse_step = 1, the current code will not work.
This commit is contained in:
Ahmed Elnaggar 2024-01-17 22:26:21 +01:00 committed by GitHub
parent fa6d12f74f
commit 98dda8ed03
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/transformers/models/switch_transformers/modeling_switch_transformers.py
View File

@ -898,7 +898,7 @@ class SwitchTransformersStack(SwitchTransformersPreTrainedModel):
config.num_layers = config.num_decoder_layers if self.is_decoder else config.num_layers
self.block = nn.ModuleList()
for i in range(config.num_layers):
is_sparse = (i % sparse_step == 1) if sparse_step > 0 else False
is_sparse = (i % sparse_step == 1 or sparse_step == 1) if sparse_step > 0 else False
self.block.append(
SwitchTransformersBlock(config, has_relative_attention_bias=bool(i == 0), is_sparse=is_sparse)