config.json

{
  "_name_or_path": "/home/jhshin/gbst-base-ds6x-newblock-972k-240517/",
  "architectures": [
    "GBSWT5ForConditionalGeneration"
  ],
  "auto_map": {
    "AutoConfig": "configuration_gbswt5.GBSWT5Config",
    "AutoModel": "modeling_gbswt5.GBSWT5ForModel",
    "AutoModelForSeq2SeqLM": "modeling_gbswt5.GBSWT5ForConditionalGeneration"
  },
  "d_ff": 3968,
  "d_kv": 64,
  "d_model": 1536,
  "decoder_start_token_id": 0,
  "dense_act_fn": "gelu_new",
  "downsample_factor": 6,
  "dropout_rate": 0.0,
  "eos_token_id": 1,
  "feed_forward_proj": "gated-gelu",
  "gbst_batchnorm": false,
  "gradient_checkpointing": false,
  "initializer_factor": 0.05,
  "is_encoder_decoder": true,
  "is_gated_act": true,
  "kv_heads": null,
  "layer_norm_epsilon": 1e-06,
  "max_subword_block_size": null,
  "model_type": "gbswt5",
  "num_decoder_layers": 6,
  "num_heads": 12,
  "num_layers": 18,
  "output_past": true,
  "pad_token_id": 0,
  "relative_attention_max_distance": 128,
  "relative_attention_num_buckets": 32,
  "score_consensus_attn": true,
  "subword_blocks": [
    [
      1,
      0
    ],
    [
      2,
      0
    ],
    [
      3,
      0
    ],
    [
      4,
      0
    ],
    [
      5,
      0
    ],
    [
      6,
      0
    ],
    [
      7,
      0
    ],
    [
      8,
      0
    ],
    [
      9,
      0
    ]
  ],
  "tie_word_embeddings": false,
  "tokenizer_class": "ByT5Tokenizer",
  "torch_dtype": "float32",
  "transformers_version": "4.33.2",
  "use_cache": true,
  "vocab_size": 384,
  "z_loss": 0.0001
}