{
  "_name_or_path": "/basemodels/models/pythia-410m",
  "abl_setting": null,
  "add_GT_ratio": 0,
  "architectures": [
    "HiarachicalPythia"
  ],
  "attention_bias": true,
  "attention_dropout": 0.0,
  "attn_implementation": "flash_attention_2",
  "block_size": 2048,
  "bos_token_id": 0,
  "chunk_merge_method": "meanpooling",
  "chunk_size": 4,
  "classifier_dropout": 0.1,
  "decoder_layers": 23,
  "detach_or_not": "pre_post",
  "dist_topk": 64,
  "distribution_merge": true,
  "encoder_layers": 1,
  "eos_token_id": 0,
  "ex_hlm_loss_ratio": 1.0,
  "ex_loss_ratio": 1.0,
  "ex_vq_loss_display": 1.0,
  "base_model_name_or_path": "/basemodels/models/pythia-410m",
  "head_factor": 1.0,
  "hidden_act": "gelu",
  "hidden_dropout": 0.0,
  "hidden_size": 1024,
  "hlm_hs_lambda": 0.5,
  "hlm_type": "vq_gpt2_ce",
  "initializer_range": 0.02,
  "intermediate_size": 4096,
  "layer_norm_eps": 1e-05,
  "layer_norm_option": "rawadd",
  "loss_abl_method": "loss2-module13-loss3-module1234",
  "loss_type": "hlm_MSE_loss",
  "max_position_embeddings": 2048,
  "model_name_or_path": "/basemodels/models/pythia-410m",
  "model_type": "gpt_neox",
  "num_attention_heads": 16,
  "num_hidden_layers": 24,
  "padding_factor": 1,
  "projector_pos": "no_projector",
  "rope_scaling": null,
  "rotary_emb_base": 10000,
  "rotary_pct": 0.25,
  "shift_feature": true,
  "special_layers": 2,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "training_type": "full_from_scratch_addvqfeature",
  "transformers_version": "4.44.0",
  "use_cache": true,
  "use_parallel_residual": true,
  "vae_config_path": "/conf/SimVQ_cb64.yaml",
  "vocab_size": 50304,
  "vq_config": {
    "codebook_size": 64,
    "kmeans_init": false,
    "learnable_codebook": false,
    "vq_type": "SimVQ"
  },
  "vq_patch_ratio": 1
}