Skip to content

Commit 87199de

Browse files
patrickvonplatenLysandreJik
authored andcommitted
fix mt5 config (#8832)
1 parent 6887947 commit 87199de

File tree

1 file changed

+4
-0
lines changed

1 file changed

+4
-0
lines changed

src/transformers/models/mt5/configuration_mt5.py

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -60,6 +60,8 @@ class MT5Config(PretrainedConfig):
6060
testing).
6161
feed_forward_proj (:obj:`string`, `optional`, defaults to :obj:`"gated-gelu"`):
6262
Type of feed forward layer to be used. Should be one of :obj:`"relu"` or :obj:`"gated-gelu"`.
63+
use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
64+
Whether or not the model should return the last key/values attentions (not used by all models).
6365
"""
6466
model_type = "mt5"
6567
keys_to_ignore_at_inference = ["past_key_values"]
@@ -79,6 +81,7 @@ def __init__(
7981
initializer_factor=1.0,
8082
feed_forward_proj="gated-gelu",
8183
is_encoder_decoder=True,
84+
use_cache=True,
8285
tokenizer_class="T5Tokenizer",
8386
tie_word_embeddings=False,
8487
pad_token_id=0,
@@ -109,6 +112,7 @@ def __init__(
109112
self.layer_norm_epsilon = layer_norm_epsilon
110113
self.initializer_factor = initializer_factor
111114
self.feed_forward_proj = feed_forward_proj
115+
self.use_cache = use_cache
112116

113117
@property
114118
def hidden_size(self):

0 commit comments

Comments
 (0)