token4m_args.json

{
  "token_roots": "/home/cuimr/csfm_data/processed/token_shards/code15_ecg_dac_rvq_last",
  "out_dir": "/home/cuimr/csfm_data/runs/physio4m_ecg_tokens_code15_2gpu",
  "variant": "small",
  "epochs": 100,
  "batch_size": 16,
  "num_workers": 4,
  "cache_size": 64,
  "lr": 0.0001,
  "weight_decay": 0.05,
  "grad_clip": 1.0,
  "val_frac": 0.05,
  "limit": null,
  "seed": 42,
  "codebook_size": 256,
  "num_quantizers": 3,
  "max_positions": 128,
  "target_token_ratio": 0.5,
  "input_token_ratio": 0.5,
  "target_channel_ratio": 0.25,
  "max_input_tokens": 256,
  "max_target_tokens": 256,
  "max_train_batches": null,
  "max_val_batches": null,
  "resume": null,
  "resume_optimizer": false,
  "ddp": true
}