_target_: text_recognizer.networks.ConvTransformer
input_dims: [1, 1, 576, 640]
hidden_dim: &hidden_dim 128
num_classes: 58
pad_index: 3
encoder:
  _target_: text_recognizer.networks.convnext.ConvNext
  dim: 16
  dim_mults: [2, 4, 8]
  depths: [3, 3, 6]
  downsampling_factors: [[2, 2], [2, 2], [2, 2]]
decoder:
  _target_: text_recognizer.networks.transformer.Decoder
  dim: *hidden_dim
  depth: 10
  block:
    _target_: text_recognizer.networks.transformer.decoder_block.DecoderBlock
    self_attn:
      _target_: text_recognizer.networks.transformer.Attention
      dim: *hidden_dim
      num_heads: 12
      dim_head: 64
      dropout_rate: &dropout_rate 0.2
      causal: true
      rotary_embedding:
        _target_: text_recognizer.networks.transformer.RotaryEmbedding
        dim: 64
    cross_attn:
      _target_: text_recognizer.networks.transformer.Attention
      dim: *hidden_dim
      num_heads: 12
      dim_head: 64
      dropout_rate: *dropout_rate
      causal: false
    norm:
      _target_: text_recognizer.networks.transformer.RMSNorm
      dim: *hidden_dim
    ff:
      _target_: text_recognizer.networks.transformer.FeedForward
      dim: *hidden_dim
      dim_out: null
      expansion_factor: 2
      glu: true
      dropout_rate: *dropout_rate
pixel_embedding:
  _target_: "text_recognizer.networks.transformer.embeddings.axial.\
    AxialPositionalEmbeddingImage"
  dim: *hidden_dim
  axial_shape: [7, 128]
  axial_dims: [64, 64]
token_pos_embedding:
  _target_: "text_recognizer.networks.transformer.embeddings.fourier.\
    PositionalEncoding"
  dim: *hidden_dim
  dropout_rate: 0.1
  max_len: 89