config.json

{
  "architectures": [
    "E1ForMaskedLM"
  ],
  "attn_backend": "sdpa",
  "auto_map": {
    "AutoConfig": "modeling_e1.E1Config",
    "AutoModel": "modeling_e1.E1Model",
    "AutoModelForMaskedLM": "modeling_e1.E1ForMaskedLM",
    "AutoModelForSequenceClassification": "modeling_e1.E1ForSequenceClassification",
    "AutoModelForTokenClassification": "modeling_e1.E1ForTokenClassification"
  },
  "bos_token_id": 1,
  "clip_qkv": 8,
  "dtype": "bfloat16",
  "eos_token_id": 2,
  "gated_mlp": true,
  "global_attention_every_n_layers": 3,
  "gradient_checkpointing": false,
  "hidden_act": "silu",
  "hidden_size": 768,
  "initializer_range": 0.02,
  "intermediate_size": 2304,
  "max_num_positions_global": 65536,
  "max_num_positions_within_seq": 8192,
  "max_num_sequences": 512,
  "model_type": "E1",
  "no_ffn_gradient_checkpointing": true,
  "num_attention_heads": 12,
  "num_hidden_layers": 20,
  "num_key_value_heads": 12,
  "pad_token_id": 0,
  "rms_norm_eps": 1e-05,
  "rope_theta_global": 500000.0,
  "rope_theta_within_seq": 10000.0,
  "tie_word_embeddings": false,
  "transformers_version": "4.57.6",
  "use_cache": true,
  "vocab_size": 34
}