_target_: text_recognizer.network.convformer.Convformer
image_height: 7
image_width: 128
patch_height: 1
patch_width: 1
dim: &dim 512
num_classes: &num_classes 57
encoder:
  _target_: text_recognizer.network.convnext.convnext.ConvNext
  dim: 16
  dim_mults: [2, 8, 32]
  depths: [2, 2, 2]
  attn: null
decoder:
  _target_: text_recognizer.network.transformer.decoder.Decoder
  dim: *dim
  ff_mult: 4
  heads: 12
  dim_head: 64
  depth: 6
  dropout_rate: 0.
  one_kv_head: true
token_embedding:
  _target_: "text_recognizer.network.transformer.embedding.token.\
    TokenEmbedding"
  num_tokens: *num_classes
  dim: *dim
  use_l2: true
tie_embeddings: false
pad_index: 3
channels: 512