Skip to content

add parameter use_key_value_cache to BartAsPointerNetworkConfig #140

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ArneBinder wants to merge 1 commit into
base: main
Choose a base branch
from
Open

add parameter use_key_value_cache to BartAsPointerNetworkConfig #140

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 12 additions & 1 deletion src/pie_modules/models/base_models/bart_as_pointer_network.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
from transformers import BartConfig, BartModel, BartPreTrainedModel
from transformers.modeling_outputs import Seq2SeqLMOutput, Seq2SeqModelOutput
from transformers.models.bart.modeling_bart import shift_tokens_right
from transformers.utils import logging
from transformers.utils import ModelOutput, logging

from pie_modules.models.base_models.bart_with_decoder_position_ids import (
BartModelWithDecoderPositionIds,
Expand Down Expand Up @@ -58,6 +58,7 @@ def __init__(
shared_decay: Optional[float] = None,
encoder_layer_norm_decay: Optional[float] = 0.001,
decoder_layer_norm_decay: Optional[float] = None,
use_key_value_cache: bool = True,
# other BartConfig parameters
**kwargs,
):
Expand All @@ -82,6 +83,8 @@ def __init__(
self.encoder_layer_norm_decay = encoder_layer_norm_decay
self.decoder_layer_norm_decay = decoder_layer_norm_decay

self.use_key_value_cache = use_key_value_cache


class BartAsPointerNetwork(BartPreTrainedModel):
config_class = BartAsPointerNetworkConfig
Expand Down Expand Up @@ -311,6 +314,14 @@ def forward(
encoder_attentions=outputs.encoder_attentions,
)

def _extract_past_from_model_output(
self, outputs: ModelOutput, standardize_cache_format: bool = False
):
if not self.config.use_key_value_cache:
return None
else:
return super()._extract_past_from_model_output(outputs, standardize_cache_format)

def prepare_inputs_for_generation(
self,
decoder_input_ids,
Expand Down