args.json

{
  "dataset_name": "c4",
  "max_seq_len": 256,
  "enc_dec_model": "facebook/bart-base",
  "train_batch_size": 32,
  "eval_batch_size": 16,
  "num_encoder_latents": 128,
  "num_decoder_latents": 128,
  "dim_ae": 768,
  "num_layers": 3,
  "l2_normalize_latents": true,
  "output_dir": "saved_latent_models/c4/2024-09-22_22-29-54",
  "save_dir": "saved_latent_models",
  "learning_rate": 0.0001,
  "num_train_steps": 50000,
  "lr_schedule": "linear",
  "lr_warmup_steps": 1000,
  "optimizer": "adamw",
  "adam_beta1": 0.9,
  "adam_beta2": 0.999,
  "adam_weight_decay": 0.01,
  "eval_every": 1000,
  "mixed_precision": "no",
  "wandb_name": "bart-c4-l2norm-128-768",
  "lm_mode": "freeze",
  "eval": false,
  "resume_training": false,
  "resume_dir": null,
  "num_devices": 1
}