config.json

{
  "architectures": [
    "FLASHQuadForMaskedLM"
  ],
  "dropout": 0.0,
  "expansion_factor": 2,
  "gradient_checkpointing": false,
  "hidden_act": "silu",
  "hidden_size": 768,
  "initializer_range": 0.02,
  "layer_norm_eps": 1e-05,
  "max_position_embeddings": 512,
  "model_type": "flash_quad",
  "norm_type": "scale_norm",
  "num_hidden_layers": 12,
  "pad_token_id": 0,
  "s": 128,
  "torch_dtype": "float32",
  "transformers_version": "4.16.2",
  "type_vocab_size": 2,
  "vocab_size": 12000
}