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

[Use cache] Align logic of use_cache with output_attentions and output_hidden_states (#5194)

Browse Source 
* fix use cache

* add bart use cache

* fix bart

* finish bart
This commit is contained in:
Patrick von Platen
2020-06-24 16:09:17 +02:00
committed by GitHub
parent 64c393ee74
commit c2a26ec8a6
13 changed files with 90 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
src/transformers/modeling_ctrl.py
View File

@@ -335,7 +335,7 @@ class CTRLModel(CTRLPreTrainedModel):
position_ids=None,
head_mask=None,
inputs_embeds=None,
use_cache=True,
use_cache=None,
output_attentions=None,
output_hidden_states=None,
):
@@ -374,6 +374,7 @@ class CTRLModel(CTRLPreTrainedModel):
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
use_cache = use_cache if use_cache is not None else self.config.use_cache
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
@@ -519,7 +520,7 @@ class CTRLLMHeadModel(CTRLPreTrainedModel):
head_mask=None,
inputs_embeds=None,
labels=None,
use_cache=True,
use_cache=None,
output_attentions=None,
output_hidden_states=None,
):