From 9ef2857c2d24d9c0a8fba3c5db58c7303124c79b Mon Sep 17 00:00:00 2001 From: Gustaf Rydholm Date: Sat, 18 Sep 2021 17:43:23 +0200 Subject: Update experiment configs --- training/conf/network/conv_transformer.yaml | 1 + training/conf/network/decoder/vae_decoder.yaml | 5 +++-- training/conf/network/encoder/vae_encoder.yaml | 5 +++-- training/conf/network/vqvae.yaml | 6 +++--- 4 files changed, 10 insertions(+), 7 deletions(-) (limited to 'training/conf/network') diff --git a/training/conf/network/conv_transformer.yaml b/training/conf/network/conv_transformer.yaml index d3a3b0f..1d61129 100644 --- a/training/conf/network/conv_transformer.yaml +++ b/training/conf/network/conv_transformer.yaml @@ -5,6 +5,7 @@ defaults: _target_: text_recognizer.networks.conv_transformer.ConvTransformer input_dims: [1, 576, 640] hidden_dim: 128 +encoder_dim: 1280 dropout_rate: 0.2 num_classes: 1006 pad_index: 1002 diff --git a/training/conf/network/decoder/vae_decoder.yaml b/training/conf/network/decoder/vae_decoder.yaml index 60cdcf1..2053544 100644 --- a/training/conf/network/decoder/vae_decoder.yaml +++ b/training/conf/network/decoder/vae_decoder.yaml @@ -1,5 +1,6 @@ _target_: text_recognizer.networks.vqvae.decoder.Decoder out_channels: 1 -hidden_dim: 64 -channels_multipliers: [8, 4, 2, 1] +hidden_dim: 32 +channels_multipliers: [4, 2, 1] dropout_rate: 0.0 +activation: leaky_relu diff --git a/training/conf/network/encoder/vae_encoder.yaml b/training/conf/network/encoder/vae_encoder.yaml index 73529fc..2ea3adf 100644 --- a/training/conf/network/encoder/vae_encoder.yaml +++ b/training/conf/network/encoder/vae_encoder.yaml @@ -1,5 +1,6 @@ _target_: text_recognizer.networks.vqvae.encoder.Encoder in_channels: 1 -hidden_dim: 64 -channels_multipliers: [1, 2, 4, 8] +hidden_dim: 32 +channels_multipliers: [1, 2, 4] dropout_rate: 0.0 +activation: leaky_relu diff --git a/training/conf/network/vqvae.yaml b/training/conf/network/vqvae.yaml index 70d27d7..d97e9b6 100644 --- a/training/conf/network/vqvae.yaml +++ b/training/conf/network/vqvae.yaml @@ -3,7 +3,7 @@ defaults: - decoder: vae_decoder _target_: text_recognizer.networks.vqvae.vqvae.VQVAE -hidden_dim: 512 -embedding_dim: 64 -num_embeddings: 4096 +hidden_dim: 128 +embedding_dim: 32 +num_embeddings: 1024 decay: 0.99 -- cgit v1.2.3-70-g09d2