config.json

{
  "activation_function": "swiglu",
  "alt_mixer": {
    "_target_": "aicl.model.models.mixers.linear_attention.LinearAttention",
    "feature_dim": 16,
    "feature_map": {
      "_target_": "aicl.model.models.mixers.linear_attention.TaylorExp",
      "input_dim": 16
    },
    "l_max": 2048,
    "num_heads": 12
  },
  "alt_mixer_layers": [
    1,
    3,
    5,
    7,
    9,
    11,
    13,
    15,
    17
  ],
  "attn_pdrop": 0,
  "bos_token_id": 50256,
  "embd_pdrop": 0,
  "eos_token_id": 50256,
  "fused_bias_fc": true,
  "fused_dropout_add_ln": true,
  "fused_mlp": false,
  "initializer_range": 0.02,
  "layer_norm_epsilon": 1e-05,
  "mixer": {
    "_target_": "aicl.model.models.mixers.convolution.BaseConv",
    "expand_proj": 4,
    "kernel_sizes": 3,
    "l_max": 2048,
    "use_bias": true
  },
  "mlp_fc1_bias": false,
  "mlp_fc2_bias": false,
  "model_type": "gpt2",
  "n_embd": 768,
  "n_head": 12,
  "n_inner": 1536,
  "n_layer": 18,
  "n_positions": 0,
  "out_proj_bias": false,
  "pad_vocab_size_multiple": 8,
  "qkv_proj_bias": false,
  "reorder_and_upcast_attn": false,
  "resid_pdrop": 0,
  "residual_in_fp32": true,
  "rms_norm": true,
  "rotary_emb_fraction": 1,
  "scale_attn_by_inverse_layer_idx": false,
  "scale_attn_weights": true,
  "special_initializer": true,
  "summary_activation": null,
  "summary_first_dropout": 0.1,
  "summary_proj_to_labels": true,
  "summary_type": "cls_index",
  "summary_use_proj": true,
  "transformers_version": "4.38.2",
  "use_cache": true,
  "use_flash_attn": true,
  "vocab_size": 50277
}