large_v3_l1.json

{
  "whisper_config": {
    "model": "large-v3",
    "layer_name": "encoder.blocks.16"
  },
  "autoencoder_variant": "l1",
  "autoencoder_config": {
    "expansion_factor": 16,
    "recon_alpha": 1e4
  },
  "seed": 0,
  "train_folder": "activations/block_16_v3_train_10k",
  "val_folder": "activations/block_16_v3_dev",
  "device": "cuda",
  "run_dir": "runs/large_v3_l1",
  "lr": 1e-4,
  "weight_decay": 0.0,
  "steps": 25000,
  "clip_thresh": 1.0,
  "batch_size": 16,
  "dl_max_workers": 0,
  "log_tb_every": 10,
  "save_every": 1000,
  "val_every": 100,
  "optimizer": "adam",
  "scheduler": "linear",
  "scheduler_params": {
    "num_warmup_steps": 1000
  },
  "start_checkpoint": null,
  "from_disk": true
}