config.json

{
  "architectures": [
    "FSTForMaskedLM"
  ],
  "bos_token_id": 101,
  "embedding_size": 1024,
  "embedding_size_f": 1024,
  "embedding_size_phi": 1024,
  "eos_token_id": 102,
  "hidden_size": 1024,
  "hidden_size_f": 1024,
  "hidden_size_phi": 1024,
  "initializer_range": 0.02,
  "intermediate_size": 4096,
  "intermediate_size_f": 4096,
  "intermediate_size_phi": 4096,
  "max_position_embeddings": 1024,
  "max_seq_len": 1024,
  "model_type": "fst",
  "num_attention_heads": 16,
  "num_attention_heads_f": 16,
  "num_attention_heads_phi": 16,
  "num_hidden_layers": 24,
  "pad_token_id": 0,
  "share_f_and_phi_embedding": true,
  "torch_dtype": "float32",
  "transformers_version": "4.52.4",
  "truncate_activation_size": false,
  "use_cache": false,
  "use_causal_attention": false,
  "vocab_size": 30522
}