{ "config": { "run_name": "dec_v3", "data": "data/v4", "cache": "data/v4/dec_v14", "seed": 0, "model": { "d": 384, "layers": 6, "heads": 6, "ff": 1536, "K": 8, "dropout": 0.05 }, "train": { "steps": 105000, "batch": 512, "lr": 1.5e-05, "wd": 0.01, "warmup": 2000, "min_lr_frac": 0.05, "schedule": "wsd", "decay_frac": 0.0, "label_smoothing": 0.02, "grad_clip": 1.0, "p_noise": 0.4, "max_angle_cos": 0.99, "logn_jitter": 0.02, "p_dz": 0.3, "rewarm": 500, "log_every": 100, "eval_every": 5000, "eval_n": 1024, "eval_ce_n": 4096, "len_buckets": 16, "max_tokens": 32768 }, "encoder": "runs/v14/best.pt" }, "step": 105000, "d_lat": 256, "best": 0.8779296875, "component": "decoder", "description": "latent -> prefix tokens, on the frozen encoder_pi", "source_checkpoint": "runs/dec_v3/best.pt", "domain": [ -3.141592653589793, 3.141592653589793 ], "format": "func-jepa-1" }