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

fix checkpoint name for wav2vec2 conformer (#18994)

Browse Source 
* fix checkpoint name for wav2vec2 conformer

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar
2022-09-12 19:39:01 +02:00
committed by GitHub
parent 8a6928e28b
commit 39b5bb79d9
2 changed files with 7 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
src/transformers/models/wav2vec2_conformer/configuration_wav2vec2_conformer.py
View File

@@ -24,8 +24,8 @@ from ...utils import logging
logger = logging.get_logger(__name__) logger = logging.get_logger(__name__)
WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP = { WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"facebook/wav2vec2-conformer-large-rel-pos": ( "facebook/wav2vec2-conformer-rel-pos-large": (
"https://huggingface.co/facebook/wav2vec2-conformer-large-rel-pos/resolve/main/config.json" "https://huggingface.co/facebook/wav2vec2-conformer-rel-pos-large/resolve/main/config.json"
), ),
} }
@@ -35,7 +35,7 @@ class Wav2Vec2ConformerConfig(PretrainedConfig):
This is the configuration class to store the configuration of a [`Wav2Vec2ConformerModel`]. It is used to This is the configuration class to store the configuration of a [`Wav2Vec2ConformerModel`]. It is used to
instantiate an Wav2Vec2Conformer model according to the specified arguments, defining the model architecture. instantiate an Wav2Vec2Conformer model according to the specified arguments, defining the model architecture.
Instantiating a configuration with the defaults will yield a similar configuration to that of the Wav2Vec2Conformer Instantiating a configuration with the defaults will yield a similar configuration to that of the Wav2Vec2Conformer
[facebook/wav2vec2-conformer-large-rel-pos](https://huggingface.co/facebook/wav2vec2-conformer-large-rel-pos) [facebook/wav2vec2-conformer-rel-pos-large](https://huggingface.co/facebook/wav2vec2-conformer-rel-pos-large)
architecture. architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
@@ -195,10 +195,10 @@ class Wav2Vec2ConformerConfig(PretrainedConfig):
```python ```python
>>> from transformers import Wav2Vec2ConformerModel, Wav2Vec2ConformerConfig >>> from transformers import Wav2Vec2ConformerModel, Wav2Vec2ConformerConfig
>>> # Initializing a Wav2Vec2Conformer facebook/wav2vec2-conformer-large-rel-pos style configuration >>> # Initializing a Wav2Vec2Conformer facebook/wav2vec2-conformer-rel-pos-large style configuration
>>> configuration = Wav2Vec2ConformerConfig() >>> configuration = Wav2Vec2ConformerConfig()
>>> # Initializing a model from the facebook/wav2vec2-conformer-large-rel-pos style configuration >>> # Initializing a model from the facebook/wav2vec2-conformer-rel-pos-large style configuration
>>> model = Wav2Vec2ConformerModel(configuration) >>> model = Wav2Vec2ConformerModel(configuration)
>>> # Accessing the model configuration >>> # Accessing the model configuration

4
src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py
View File

@@ -80,7 +80,7 @@ _XVECTOR_EXPECTED_OUTPUT = 1.0
WAV2VEC2_CONFORMER_PRETRAINED_MODEL_ARCHIVE_LIST = [ WAV2VEC2_CONFORMER_PRETRAINED_MODEL_ARCHIVE_LIST = [
"facebook/wav2vec2-conformer-large-rel-pos", "facebook/wav2vec2-conformer-rel-pos-large",
# See all Wav2Vec2Conformer models at https://huggingface.co/models?filter=wav2vec2-conformer # See all Wav2Vec2Conformer models at https://huggingface.co/models?filter=wav2vec2-conformer
] ]
@@ -1226,7 +1226,7 @@ WAV2VEC2_CONFORMER_INPUTS_DOCSTRING = r"""
`attention_mask` should only be passed if the corresponding processor has `config.return_attention_mask == `attention_mask` should only be passed if the corresponding processor has `config.return_attention_mask ==
True`. For all models whose processor has `config.return_attention_mask == False`, such as True`. For all models whose processor has `config.return_attention_mask == False`, such as
[wav2vec2_conformer-base](https://huggingface.co/facebook/wav2vec2-conformer-large-rel-pos), [wav2vec2-conformer-rel-pos-large](https://huggingface.co/facebook/wav2vec2-conformer-rel-pos-large),
`attention_mask` should **not** be passed to avoid degraded performance when doing batched inference. For `attention_mask` should **not** be passed to avoid degraded performance when doing batched inference. For
such models `input_values` should simply be padded with 0 and passed without `attention_mask`. Be aware such models `input_values` should simply be padded with 0 and passed without `attention_mask`. Be aware
that these models also yield slightly different results depending on whether `input_values` is padded or that these models also yield slightly different results depending on whether `input_values` is padded or