config.json

{
  "base_model": "google/paligemma2-3b-pt-448",
  "ft_model": "InAbsentia/cua-agent-run-k-cot-probe",
  "data_path": "data/agentnet_sft_cot/train_with_images.jsonl",
  "images_root": "data/agentnet_sft_cot/",
  "layers": [
    21,
    22,
    23,
    24
  ],
  "expansion": 48,
  "model_family": "dfc",
  "activation": "jumprelu",
  "k": 75,
  "threshold_init": 0.003,
  "bandwidth": 1.0,
  "steps": 50000,
  "batch_size": 320,
  "lr": 0.0002,
  "sparsity_coeff": 0.0005,
  "tanh_c": 1.5,
  "pre_act_coeff": 3e-05,
  "max_grad_norm": 1.0,
  "auxk_coeff": 0.03125,
  "ramp_steps": 3500,
  "checkpoint_every": 5000,
  "skip_collection": true,
  "skip_training": false,
  "skip_analysis": true,
  "resume": "runs/cross_model_clt_dfc_v1/checkpoints/step_006000.pt",
  "output_dir": "runs/cross_model_clt_dfc_v1",
  "mixed_precision": "bf16",
  "wandb_project": null,
  "compile": true,
  "model_metadata": {
    "model_family": "dfc",
    "activation": "jumprelu",
    "partition_ratios": {
      "shared": 0.5,
      "base_exclusive": 0.25,
      "ft_exclusive": 0.25
    },
    "partition_dims": {
      "shared": 55296,
      "base_exclusive": 27648,
      "ft_exclusive": 27648
    }
  }
}