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

Fix cross-attention head mask for Torch encoder-decoder models (#10605)

Browse Source 
* Fix cross-attention head mask for Torch BART models

* Fix head masking for cross-attention module for the following
models: BART, Blenderbot, Blenderbot_small, M2M_100, Marian, MBart,
Pegasus

* Enable test_headmasking for M2M_100 model

* Fix cross_head_mask for FSMT, LED and T5

* This commit fixes `head_mask` for cross-attention modules
in the following models: FSMT, LED, T5

* It also contains some smaller changes in doc so that
it is be perfectly clear the shape of `cross_head_mask`
is the same as of `decoder_head_mask`

* Update template

* Fix template for BartForCausalLM

* Fix cross_head_mask for Speech2Text models

* Fix cross_head_mask in templates

* Fix args order in BartForCausalLM template

* Fix doc in BART templates

* Make more explicit naming

* `cross_head_mask` -> `cross_attn_head_mask`

* `cross_layer_head_mask` -> `cross_attn_layer_head_mask`

* Fix doc

* make style quality

* Fix speech2text docstring
This commit is contained in:
Daniel Stancl
2021-04-23 18:58:06 +02:00
committed by GitHub
parent ca6b80cadb
commit e3ff165aa5
23 changed files with 587 additions and 389 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/test_modeling_pegasus.py
View File

@@ -42,6 +42,7 @@ def prepare_pegasus_inputs_dict(
decoder_attention_mask=None,
head_mask=None,
decoder_head_mask=None,
cross_attn_head_mask=None,
):
if attention_mask is None:
attention_mask = input_ids.ne(config.pad_token_id)
@@ -51,6 +52,8 @@ def prepare_pegasus_inputs_dict(
head_mask = torch.ones(config.encoder_layers, config.encoder_attention_heads, device=torch_device)
if decoder_head_mask is None:
decoder_head_mask = torch.ones(config.decoder_layers, config.decoder_attention_heads, device=torch_device)
if cross_attn_head_mask is None:
cross_attn_head_mask = torch.ones(config.decoder_layers, config.decoder_attention_heads, device=torch_device)
return {
"input_ids": input_ids,
"decoder_input_ids": decoder_input_ids,
@@ -58,6 +61,7 @@ def prepare_pegasus_inputs_dict(
"decoder_attention_mask": attention_mask,
"head_mask": head_mask,
"decoder_head_mask": decoder_head_mask,
"cross_attn_head_mask": cross_attn_head_mask,
}