{
  "_name_or_path": "/nlp/scr3/nlp/llms-in-llms/mrt5/models/span_corruption_multilingual/MrT5/mrt5-large_span_corruption_multilingual_50%_seed23/checkpoints/checkpoint-5000/",
  "architectures": [
    "MrT5ForConditionalGeneration"
  ],
  "auto_map": {
    "AutoConfig": "configuration_mrt5.MrT5Config",
    "AutoModelForSeq2SeqLM": "modeling_mrt5.MrT5ForConditionalGeneration"
  },
  "classifier_dropout": 0.0,
  "d_ff": 3840,
  "d_kv": 64,
  "d_model": 1536,
  "decoder_start_token_id": 0,
  "delete_gate_layer": 3,
  "deletion_threshold": -15.0,
  "deletion_type": "scaled_sigmoid",
  "dense_act_fn": "gelu_new",
  "dropout_rate": 0.1,
  "eos_token_id": 1,
  "eval_language": "en",
  "feed_forward_proj": "gated-gelu",
  "fixed_deletion_amount": 0.5,
  "gate_layer_norm": true,
  "gradient_checkpointing": false,
  "has_absolute_position_embeddings": false,
  "initializer_factor": 1.0,
  "is_encoder_decoder": true,
  "is_gated_act": true,
  "layer_norm_epsilon": 1e-06,
  "model_type": "mrt5",
  "num_decoder_layers": 12,
  "num_heads": 16,
  "num_layers": 36,
  "output_past": true,
  "pad_token_id": 0,
  "random_deletion_probability": 0.5,
  "relative_attention_max_distance": 128,
  "relative_attention_num_buckets": 32,
  "sigmoid_mask_scale": -30.0,
  "tie_word_embeddings": false,
  "tokenizer_class": "ByT5Tokenizer",
  "torch_dtype": "float32",
  "train_language": "en",
  "transformers_version": "4.39.1",
  "use_cache": true,
  "use_gumbel_noise": false,
  "use_softmax1": true,
  "vocab_size": 384
}