summaryrefslogtreecommitdiff
path: root/training/conf/network/vit_lines.yaml
diff options
context:
space:
mode:
Diffstat (limited to 'training/conf/network/vit_lines.yaml')
-rw-r--r--training/conf/network/vit_lines.yaml14
1 files changed, 7 insertions, 7 deletions
diff --git a/training/conf/network/vit_lines.yaml b/training/conf/network/vit_lines.yaml
index 35f83c3..f32cb83 100644
--- a/training/conf/network/vit_lines.yaml
+++ b/training/conf/network/vit_lines.yaml
@@ -3,21 +3,21 @@ image_height: 56
image_width: 1024
patch_height: 28
patch_width: 32
-dim: &dim 256
-num_classes: &num_classes 57
+dim: &dim 1024
+num_classes: &num_classes 58
encoder:
_target_: text_recognizer.network.transformer.encoder.Encoder
dim: *dim
- inner_dim: 1024
- heads: 8
+ inner_dim: 2048
+ heads: 16
dim_head: 64
depth: 6
dropout_rate: 0.0
decoder:
_target_: text_recognizer.network.transformer.decoder.Decoder
dim: *dim
- inner_dim: 1024
- heads: 8
+ inner_dim: 2048
+ heads: 16
dim_head: 64
depth: 6
dropout_rate: 0.0
@@ -33,5 +33,5 @@ pos_embedding:
dim: *dim
max_length: 89
use_l2: true
-tie_embeddings: true
+tie_embeddings: false
pad_index: 3