3 files changed, 1 insertions, 36 deletions
diff --git a/training/conf/network/conv_transformer.yaml b/training/conf/network/conv_transformer.yaml
index 54eb028..39c5c46 100644
--- a/training/conf/network/conv_transformer.yaml
+++ b/training/conf/network/conv_transformer.yaml
@@ -10,7 +10,7 @@ encoder:
   bn_momentum: 0.99
   bn_eps: 1.0e-3
   depth: 3
-  out_channels: 128
+  out_channels: *hidden_dim
 decoder:
   _target_: text_recognizer.networks.transformer.Decoder
   depth: 6
diff --git a/training/conf/network/decoder/transformer_decoder.yaml b/training/conf/network/decoder/transformer_decoder.yaml
deleted file mode 100644
index 4588ee9..0000000
--- a/training/conf/network/decoder/transformer_decoder.yaml
+++ /dev/null
@@ -1,30 +0,0 @@
-_target_: text_recognizer.networks.transformer.decoder.Decoder
-depth: 4
-block:
-  _target_: text_recognizer.networks.transformer.decoder.DecoderBlock
-  self_attn:
-    _target_: text_recognizer.networks.transformer.attention.Attention
-    dim: 64
-    num_heads: 4
-    dim_head: 64
-    dropout_rate: 0.05
-    causal: true
-    rotary_embedding:
-      _target_: text_recognizer.networks.transformer.embeddings.rotary.RotaryEmbedding
-      dim: 128
-  cross_attn:
-    _target_: text_recognizer.networks.transformer.attention.Attention
-    dim: 64
-    num_heads: 4
-    dim_head: 64
-    dropout_rate: 0.05
-    causal: false
-  norm:
-    _target_: text_recognizer.networks.transformer.norm.RMSNorm
-    normalized_shape: 192
-  ff:
-    _target_: text_recognizer.networks.transformer.mlp.FeedForward
-    dim_out: null
-    expansion_factor: 4
-    glu: true
-    dropout_rate: 0.2
diff --git a/training/conf/network/encoder/efficientnet.yaml b/training/conf/network/encoder/efficientnet.yaml
deleted file mode 100644
index a7be069..0000000
--- a/training/conf/network/encoder/efficientnet.yaml
+++ /dev/null
@@ -1,5 +0,0 @@
-_target_: text_recognizer.networks.efficientnet.EfficientNet
-arch: b0
-stochastic_dropout_rate: 0.2
-bn_momentum: 0.99
-bn_eps: 1.0e-3