Skip to content

GitLab

Commit 39d72bcc authored by LysandreJik's avatar LysandreJik
Browse files

Fixed the RoBERTa checkpoint conversion script according to the LM head refactoring.

parent 770043ee
Show whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
pytorch_transformers/convert_roberta_checkpoint_to_pytorch.py
View file @ 39d72bcc
...@@ -123,7 +123,7 @@ def convert_roberta_checkpoint_to_pytorch(roberta_checkpoint_path, pytorch_dump_ ...@@ -123,7 +123,7 @@ def convert_roberta_checkpoint_to_pytorch(roberta_checkpoint_path, pytorch_dump_
model.lm_head.layer_norm.weight = roberta.model.decoder.lm_head.layer_norm.weight model.lm_head.layer_norm.weight = roberta.model.decoder.lm_head.layer_norm.weight
model.lm_head.layer_norm.bias = roberta.model.decoder.lm_head.layer_norm.bias model.lm_head.layer_norm.bias = roberta.model.decoder.lm_head.layer_norm.bias
model.lm_head.layer_norm.variance_epsilon = roberta.model.decoder.lm_head.layer_norm.eps model.lm_head.layer_norm.variance_epsilon = roberta.model.decoder.lm_head.layer_norm.eps
model.lm_head.weight = roberta.model.decoder.lm_head.weight model.lm_head.decoder.weight = roberta.model.decoder.lm_head.weight
model.lm_head.bias = roberta.model.decoder.lm_head.bias model.lm_head.bias = roberta.model.decoder.lm_head.bias
# Let's check that we get the same results. # Let's check that we get the same results.
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment