Skip to content

Make loading of pretrained gpt2 faster by avoiding initialization of Conv1D's weights#21879

Merged
sgugger merged 1 commit into
huggingface:mainfrom
twaka:reordering-initialization-of-conv1d
Mar 1, 2023
Merged

Make loading of pretrained gpt2 faster by avoiding initialization of Conv1D's weights#21879
sgugger merged 1 commit into
huggingface:mainfrom
twaka:reordering-initialization-of-conv1d

Commits

Commits on Mar 1, 2023