SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Add migrating from pytorch-transformers (#4273)

Browse Source 
"Migrating from pytorch-transformers to transformers" is missing in the main document. It is available in the main `readme` thought. Just move it to the document.
This commit is contained in:
Guo, Quan
2020-05-11 13:35:13 -04:00
committed by GitHub
parent 051dcb2a07
commit 39994051e4
1 changed files with 14 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
docs/source/migration.md
View File

@@ -1,5 +1,18 @@
# Migrating from pytorch-pretrained-bert
# Migrating from previous packages
## Migrating from pytorch-transformers to transformers
Here is a quick summary of what you should take care of when migrating from `pytorch-transformers` to `transformers`.
### Positional order of some models' keywords inputs (`attention_mask`, `token_type_ids`...) changed
To be able to use Torchscript (see #1010, #1204 and #1195) the specific order of some models **keywords inputs** (`attention_mask`, `token_type_ids`...) has been changed.
If you used to call the models with keyword names for keyword arguments, e.g. `model(inputs_ids, attention_mask=attention_mask, token_type_ids=token_type_ids)`, this should not cause any change.
If you used to call the models with positional inputs for keyword arguments, e.g. `model(inputs_ids, attention_mask, token_type_ids)`, you may have to double check the exact order of input arguments.
## Migrating from pytorch-pretrained-bert
Here is a quick summary of what you should take care of when migrating from `pytorch-pretrained-bert` to `transformers`