config.json

{
  "architectures": [
    "Pix2StructVisionModel"
  ],
  "attention_dropout": 0.0,
  "d_ff": 3968,
  "d_kv": 64,
  "dense_act_fn": "gelu_new",
  "dropout_rate": 0.0,
  "hidden_size": 1536,
  "initializer_factor": 1.0,
  "initializer_range": 0.02,
  "layer_norm_bias": false,
  "layer_norm_eps": 1e-06,
  "model_type": "pix2struct_vision_model",
  "num_attention_heads": 24,
  "num_channels": 3,
  "num_hidden_layers": 18,
  "patch_embed_hidden_size": 768,
  "patch_size": 16,
  "projection_dim": 768,
  "relative_attention_max_distance": 128,
  "relative_attention_num_buckets": 32,
  "seq_len": 4096,
  "torch_dtype": "float32",
  "transformers_version": "4.30.2"
}