From 3fbb55c75779824aacfc43067f0892674a9cfbc6 Mon Sep 17 00:00:00 2001 From: Bhadresh Savani Date: Wed, 15 Sep 2021 11:03:52 +0530 Subject: [PATCH] [Flax] Fixes typo in Bart based Flax Models (#13565) --- src/transformers/models/bart/modeling_flax_bart.py | 8 ++++---- src/transformers/models/marian/modeling_flax_marian.py | 8 ++++---- src/transformers/models/mbart/modeling_flax_mbart.py | 8 ++++---- src/transformers/models/pegasus/modeling_flax_pegasus.py | 8 ++++---- .../modeling_flax_{{cookiecutter.lowercase_modelname}}.py | 8 ++++---- 5 files changed, 20 insertions(+), 20 deletions(-) diff --git a/src/transformers/models/bart/modeling_flax_bart.py b/src/transformers/models/bart/modeling_flax_bart.py index 9c698b495a..d0aa80223a 100644 --- a/src/transformers/models/bart/modeling_flax_bart.py +++ b/src/transformers/models/bart/modeling_flax_bart.py @@ -406,7 +406,7 @@ class FlaxBartEncoderLayer(nn.Module): self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.fc1 = nn.Dense( self.config.encoder_ffn_dim, dtype=self.dtype, @@ -433,7 +433,7 @@ class FlaxBartEncoderLayer(nn.Module): residual = hidden_states hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states @@ -515,7 +515,7 @@ class FlaxBartDecoderLayer(nn.Module): ) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.encoder_attn = FlaxBartAttention( @@ -572,7 +572,7 @@ class FlaxBartDecoderLayer(nn.Module): # Fully Connected residual = hidden_states hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states diff --git a/src/transformers/models/marian/modeling_flax_marian.py b/src/transformers/models/marian/modeling_flax_marian.py index c90f07b2fe..a39a9f5db6 100644 --- a/src/transformers/models/marian/modeling_flax_marian.py +++ b/src/transformers/models/marian/modeling_flax_marian.py @@ -411,7 +411,7 @@ class FlaxMarianEncoderLayer(nn.Module): self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.fc1 = nn.Dense( self.config.encoder_ffn_dim, dtype=self.dtype, @@ -438,7 +438,7 @@ class FlaxMarianEncoderLayer(nn.Module): residual = hidden_states hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states @@ -523,7 +523,7 @@ class FlaxMarianDecoderLayer(nn.Module): ) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.encoder_attn = FlaxMarianAttention( @@ -580,7 +580,7 @@ class FlaxMarianDecoderLayer(nn.Module): # Fully Connected residual = hidden_states hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states diff --git a/src/transformers/models/mbart/modeling_flax_mbart.py b/src/transformers/models/mbart/modeling_flax_mbart.py index 77dc79f795..3801bad326 100644 --- a/src/transformers/models/mbart/modeling_flax_mbart.py +++ b/src/transformers/models/mbart/modeling_flax_mbart.py @@ -417,7 +417,7 @@ class FlaxMBartEncoderLayer(nn.Module): self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.fc1 = nn.Dense( self.config.encoder_ffn_dim, dtype=self.dtype, @@ -444,7 +444,7 @@ class FlaxMBartEncoderLayer(nn.Module): residual = hidden_states hidden_states = self.final_layer_norm(hidden_states) hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states @@ -527,7 +527,7 @@ class FlaxMBartDecoderLayer(nn.Module): ) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.encoder_attn = FlaxMBartAttention( @@ -585,7 +585,7 @@ class FlaxMBartDecoderLayer(nn.Module): residual = hidden_states hidden_states = self.final_layer_norm(hidden_states) hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states diff --git a/src/transformers/models/pegasus/modeling_flax_pegasus.py b/src/transformers/models/pegasus/modeling_flax_pegasus.py index bf6bdb93fb..9a181d81d2 100644 --- a/src/transformers/models/pegasus/modeling_flax_pegasus.py +++ b/src/transformers/models/pegasus/modeling_flax_pegasus.py @@ -411,7 +411,7 @@ class FlaxPegasusEncoderLayer(nn.Module): self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.fc1 = nn.Dense( self.config.encoder_ffn_dim, dtype=self.dtype, @@ -438,7 +438,7 @@ class FlaxPegasusEncoderLayer(nn.Module): residual = hidden_states hidden_states = self.final_layer_norm(hidden_states) hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states @@ -522,7 +522,7 @@ class FlaxPegasusDecoderLayer(nn.Module): ) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.encoder_attn = FlaxPegasusAttention( @@ -580,7 +580,7 @@ class FlaxPegasusDecoderLayer(nn.Module): residual = hidden_states hidden_states = self.final_layer_norm(hidden_states) hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states diff --git a/templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_flax_{{cookiecutter.lowercase_modelname}}.py b/templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_flax_{{cookiecutter.lowercase_modelname}}.py index 1692ae393c..9386129514 100644 --- a/templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_flax_{{cookiecutter.lowercase_modelname}}.py +++ b/templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/modeling_flax_{{cookiecutter.lowercase_modelname}}.py @@ -1432,7 +1432,7 @@ class Flax{{cookiecutter.camelcase_modelname}}EncoderLayer(nn.Module): self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.fc1 = nn.Dense( self.config.encoder_ffn_dim, dtype=self.dtype, @@ -1459,7 +1459,7 @@ class Flax{{cookiecutter.camelcase_modelname}}EncoderLayer(nn.Module): residual = hidden_states hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states @@ -1541,7 +1541,7 @@ class Flax{{cookiecutter.camelcase_modelname}}DecoderLayer(nn.Module): ) self.dropout_layer = nn.Dropout(rate=self.config.dropout) self.activation_fn = ACT2FN[self.config.activation_function] - self.acticvation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) + self.activation_dropout_layer = nn.Dropout(rate=self.config.activation_dropout) self.self_attn_layer_norm = nn.LayerNorm(dtype=self.dtype) self.encoder_attn = Flax{{cookiecutter.camelcase_modelname}}Attention( @@ -1598,7 +1598,7 @@ class Flax{{cookiecutter.camelcase_modelname}}DecoderLayer(nn.Module): # Fully Connected residual = hidden_states hidden_states = self.activation_fn(self.fc1(hidden_states)) - hidden_states = self.acticvation_dropout_layer(hidden_states, deterministic=deterministic) + hidden_states = self.activation_dropout_layer(hidden_states, deterministic=deterministic) hidden_states = self.fc2(hidden_states) hidden_states = self.dropout_layer(hidden_states, deterministic=deterministic) hidden_states = residual + hidden_states