config.json

{
  "architectures": [
    "BertForPreTraining"
  ],
  "attention_dropout": 0.0,
  "attention_probs_dropout_prob": 0.1,
  "bos_token_id": 151643,
  "classifier_dropout": null,
  "eos_token_id": 151643,
  "hidden_act": "silu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 1024,
  "initializer_range": 0.02,
  "intermediate_size": 2816,
  "layer_norm_eps": 1e-12,
  "max_position_embeddings": 32768,
  "max_window_layers": 21,
  "model_type": "bert",
  "num_attention_heads": 16,
  "num_hidden_layers": 24,
  "num_key_value_heads": 16,
  "pad_token_id": 0,
  "position_embedding_type": "absolute",
  "rms_norm_eps": 1e-06,
  "rope_theta": 1000000.0,
  "sliding_window": 32768,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.43.0.dev0",
  "type_vocab_size": 2,
  "use_cache": true,
  "use_sliding_window": false,
  "vocab_size": 151936
}