config.json

{
  "MoEOmniGenomefold_config": null,
  "_name_or_path": "checkpoint-214000/",
  "architectures": [
    "MoEOmniGenomeForMaskedLM"
  ],
  "attention_dropout": 0.0,
  "attention_probs_dropout_prob": 0.0,
  "auto_map": {
    "AutoConfig": "configuration_moeomnigenome.MoEOmniGenomeConfig",
    "AutoModel": "modeling_moeomnigenome.MoEOmniGenomeModel",
    "AutoModelForMaskedLM": "modeling_moeomnigenome.MoEOmniGenomeForMaskedLM",
    "AutoModelForSeq2SeqLM": "modeling_moeomnigenome.MoEOmniGenomeForSeq2SeqLM",
    "AutoModelForSequenceClassification": "modeling_moeomnigenome.MoEOmniGenomeForSequenceClassification",
    "AutoModelForTokenClassification": "modeling_moeomnigenome.MoEOmniGenomeForTokenClassification"
  },
  "bos_token_id": 1,
  "classifier_dropout": null,
  "emb_layer_norm_before": false,
  "embed_bpp": true,
  "eos_token_id": 2,
  "hidden_act": "silu",
  "hidden_dropout_prob": 0,
  "hidden_size": 600,
  "id2label": {
    "0": "(",
    "1": ")",
    "2": "."
  },
  "initializer_range": 0.02,
  "intermediate_size": 2400,
  "is_folding_model": false,
  "label2id": {
    "(": 0,
    ")": 1,
    ".": 2
  },
  "layer_norm_eps": 1e-05,
  "mask_token_id": 23,
  "max_position_embeddings": 1026,
  "model_type": "moeomnigenome",
  "num_attention_heads": 30,
  "num_experts_per_tok": 2,
  "num_generation": 50,
  "num_hidden_layers": 16,
  "num_key_value_heads": 8,
  "num_local_experts": 8,
  "num_population": 100,
  "output_router_logits": false,
  "pad_token_id": 1,
  "position_embedding_type": "rotary",
  "rms_norm_eps": 1e-05,
  "rope_theta": 1000000.0,
  "router_aux_loss_coef": 0.02,
  "router_jitter_noise": 0.01,
  "sliding_window": null,
  "tie_word_embeddings": false,
  "token_dropout": true,
  "torch_dtype": "float16",
  "transformers_version": "4.37.2",
  "use_cache": true,
  "vocab_list": null,
  "vocab_size": 64
}