diff options
| author | Gustaf Rydholm <gustaf.rydholm@gmail.com> | 2021-10-27 22:16:39 +0200 |
|---|---|---|
| committer | Gustaf Rydholm <gustaf.rydholm@gmail.com> | 2021-10-27 22:16:39 +0200 |
| commit | 3e24b92ee1bac124ea8c7bddb15236ccc5fe300d (patch) | |
| tree | 81803764977345b9264b166926024559908cb066 /training/conf/experiment/conv_transformer_paragraphs_wp.yaml | |
| parent | 4a6550ddef7d1f1971737bc22715db6381441f79 (diff) | |
Update to configs
Diffstat (limited to 'training/conf/experiment/conv_transformer_paragraphs_wp.yaml')
| -rw-r--r-- | training/conf/experiment/conv_transformer_paragraphs_wp.yaml | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/training/conf/experiment/conv_transformer_paragraphs_wp.yaml b/training/conf/experiment/conv_transformer_paragraphs_wp.yaml index ebaa17a..91fba9a 100644 --- a/training/conf/experiment/conv_transformer_paragraphs_wp.yaml +++ b/training/conf/experiment/conv_transformer_paragraphs_wp.yaml @@ -103,14 +103,14 @@ network: attn_fn: text_recognizer.networks.transformer.attention.Attention attn_kwargs: dim_head: 32 - dropout_rate: 0.2 + dropout_rate: 0.05 norm_fn: text_recognizer.networks.transformer.norm.ScaleNorm ff_fn: text_recognizer.networks.transformer.mlp.FeedForward ff_kwargs: dim_out: null expansion_factor: 4 glu: true - dropout_rate: 0.2 + dropout_rate: 0.05 cross_attend: true pre_norm: true rotary_emb: @@ -124,7 +124,7 @@ network: token_pos_embedding: _target_: text_recognizer.networks.transformer.positional_encodings.PositionalEncoding hidden_dim: *hidden_dim - dropout_rate: 0.2 + dropout_rate: 0.05 max_len: *max_output_len model: |