summaryrefslogtreecommitdiff
path: root/training/conf/network
diff options
context:
space:
mode:
Diffstat (limited to 'training/conf/network')
-rw-r--r--training/conf/network/conv_transformer.yaml3
-rw-r--r--training/conf/network/efficientnet.yaml7
2 files changed, 9 insertions, 1 deletions
diff --git a/training/conf/network/conv_transformer.yaml b/training/conf/network/conv_transformer.yaml
index 39c5c46..ccdf960 100644
--- a/training/conf/network/conv_transformer.yaml
+++ b/training/conf/network/conv_transformer.yaml
@@ -11,6 +11,7 @@ encoder:
bn_eps: 1.0e-3
depth: 3
out_channels: *hidden_dim
+ stride: [2, 1]
decoder:
_target_: text_recognizer.networks.transformer.Decoder
depth: 6
@@ -46,4 +47,4 @@ decoder:
pixel_embedding:
_target_: text_recognizer.networks.transformer.AxialPositionalEmbedding
dim: *hidden_dim
- shape: [72, 80]
+ shape: [17, 78]
diff --git a/training/conf/network/efficientnet.yaml b/training/conf/network/efficientnet.yaml
new file mode 100644
index 0000000..bbe68dd
--- /dev/null
+++ b/training/conf/network/efficientnet.yaml
@@ -0,0 +1,7 @@
+_target_: text_recognizer.networks.efficientnet.EfficientNet
+arch: b0
+stochastic_dropout_rate: 0.2
+bn_momentum: 0.99
+bn_eps: 1.0e-3
+stride: [2, 1]
+out_channels: 144