Upload weights/training_args.json
Browse files- weights/training_args.json +77 -0
weights/training_args.json
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dataset_name": "MultimodalUniverse/desi",
|
| 3 |
+
"max_samples": 97332,
|
| 4 |
+
"cache_dir": "/workspace/native_specz_mae/cache",
|
| 5 |
+
"hf_cache_dir": "/workspace/hf_cache/datasets",
|
| 6 |
+
"run_dir": "/workspace/runs/conv300m_97k_wavegrid_gentle",
|
| 7 |
+
"resume_checkpoint": "/workspace/runs/hybrid_specz_300m_oodprobe_joint/20260518_035753/best.pt",
|
| 8 |
+
"allow_mismatched_checkpoint": false,
|
| 9 |
+
"refresh_data": false,
|
| 10 |
+
"epochs": 3,
|
| 11 |
+
"batch_size": 8,
|
| 12 |
+
"num_workers": 4,
|
| 13 |
+
"target_length": 8192,
|
| 14 |
+
"architecture": "conv",
|
| 15 |
+
"d_model": 1024,
|
| 16 |
+
"conv_width": 512,
|
| 17 |
+
"layers": 20,
|
| 18 |
+
"heads": 16,
|
| 19 |
+
"dropout": 0.1,
|
| 20 |
+
"z_bins": 64,
|
| 21 |
+
"stem_stride": 8,
|
| 22 |
+
"token_stride": 8,
|
| 23 |
+
"rec_hidden_mult": 0,
|
| 24 |
+
"rec_refine_width": 16,
|
| 25 |
+
"rec_refine_kernel": 5,
|
| 26 |
+
"layerscale_init": 0.0,
|
| 27 |
+
"prediction_mode": "bin_residual",
|
| 28 |
+
"bin_temperature": 1.0,
|
| 29 |
+
"residual_scale": 0.06,
|
| 30 |
+
"candidate_topk": 5,
|
| 31 |
+
"lr": 2e-06,
|
| 32 |
+
"min_lr": 5e-07,
|
| 33 |
+
"warmup_steps": 50,
|
| 34 |
+
"weight_decay": 0.03,
|
| 35 |
+
"grad_clip": 1.0,
|
| 36 |
+
"grad_accum_steps": 1,
|
| 37 |
+
"eval_every": 250,
|
| 38 |
+
"eval_max_val": 1200,
|
| 39 |
+
"eval_max_ood": 700,
|
| 40 |
+
"max_steps": 1500,
|
| 41 |
+
"checkpoint_score": "combined",
|
| 42 |
+
"score_z_alpha": 0.7,
|
| 43 |
+
"objective": "joint",
|
| 44 |
+
"balance_redshift": false,
|
| 45 |
+
"train_clean_only": false,
|
| 46 |
+
"clean_sample_boost": 3.0,
|
| 47 |
+
"augment_ood": true,
|
| 48 |
+
"eval_ood": true,
|
| 49 |
+
"random_mask_ratio": 0.3,
|
| 50 |
+
"eval_mask_ratio": 0.3,
|
| 51 |
+
"mask_mode": "mixed_span",
|
| 52 |
+
"mask_span_min": 16,
|
| 53 |
+
"mask_span_max": 96,
|
| 54 |
+
"line_region_percentile": 90.0,
|
| 55 |
+
"crop_prob": 0.25,
|
| 56 |
+
"bad_window_prob": 0.18,
|
| 57 |
+
"throughput_prob": 0.35,
|
| 58 |
+
"noise_prob": 0.2,
|
| 59 |
+
"resolution_prob": 0.25,
|
| 60 |
+
"downsample_prob": 0.25,
|
| 61 |
+
"line_dropout_prob": 0.12,
|
| 62 |
+
"span_dropout_prob": 0.18,
|
| 63 |
+
"grid_jitter_prob": 0.35,
|
| 64 |
+
"grid_shift_frac": 0.015,
|
| 65 |
+
"grid_scale_frac": 0.04,
|
| 66 |
+
"redshift_shift": 0.0,
|
| 67 |
+
"rec_weight": 0.4,
|
| 68 |
+
"z_weight": 1.0,
|
| 69 |
+
"z_bin_weight": 0.35,
|
| 70 |
+
"z_candidate_weight": 0.1,
|
| 71 |
+
"z_nll_weight": 0.02,
|
| 72 |
+
"zwarn_weight": 0.3,
|
| 73 |
+
"high_z_boost": 2.0,
|
| 74 |
+
"high_z_threshold": 1.0,
|
| 75 |
+
"clean_z_only": false,
|
| 76 |
+
"seed": 17
|
| 77 |
+
}
|