tied_per_target_skip_global_batchtopk_jumprelu/cfg.json

{
    "num_features": 122880,
    "num_layers": 12,
    "d_model": 768,
    "model_name": null,
    "normalization_method": "mean_std",
    "activation_fn": "jumprelu",
    "jumprelu_threshold": 0.0,
    "batchtopk_k": null,
    "batchtopk_straight_through": false,
    "topk_k": null,
    "topk_straight_through": true,
    "topk_mode": "global",
    "two_stage_batchtopk": false,
    "two_stage_topk": false,
    "clt_dtype": null,
    "expected_input_dtype": null,
    "mlp_input_template": null,
    "mlp_output_template": null,
    "tl_input_template": null,
    "tl_output_template": null,
    "decoder_tying": "per_target",
    "enable_feature_offset": false,
    "enable_feature_scale": false,
    "skip_connection": true
}