{
  "_name_or_path": "pszemraj/mega-small-embed-syntheticSTS-16384",
  "activation": "silu",
  "add_lm_hidden_dense_layer": false,
  "add_token_type_embeddings": true,
  "architectures": [
    "MegaModel"
  ],
  "attention_activation": "softmax",
  "attention_probs_dropout_prob": 0,
  "bidirectional": true,
  "bos_token_id": 0,
  "chunk_size": 1024,
  "classifier_dropout": null,
  "dropout_prob": 0.05,
  "ema_beta_range": 0.02,
  "ema_delta_alpha_range": 0.2,
  "ema_gamma_omega_range": 1.0,
  "ema_projection_size": 32,
  "eos_token_id": 2,
  "hidden_dropout_prob": 0,
  "hidden_size": 768,
  "initializer_range": 0.02,
  "intermediate_size": 2304,
  "max_positions": 16384,
  "model_type": "mega",
  "nffn_activation_dropout_prob": 0,
  "nffn_hidden_size": 2304,
  "norm_affine": true,
  "normalization_type": "scalenorm",
  "normalize_before_ffn": false,
  "normalize_before_mega": false,
  "num_attention_heads": 1,
  "num_hidden_layers": 8,
  "pad_token_id": 1,
  "relative_positional_bias": "simple",
  "sep_token_id": 2,
  "shared_representation_size": 192,
  "torch_dtype": "float32",
  "transformers_version": "4.38.2",
  "truncation": null,
  "type_vocab_size": 2,
  "use_cache": true,
  "use_chunking": true,
  "use_feature_dropout": false,
  "use_normalized_ffn": true,
  "vocab_size": 50304
}