Upload training_args.json
Browse files- training_args.json +87 -0
training_args.json
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cmdline": "birder/scripts/train_i_jepa.py --network vit_reg4_m16_rms_avg --predictor-depth 6 --opt adamw --lr 0.001 --lr-scheduler cosine --lr-cosine-min 1e-6 --epochs 60 --warmup-epochs 5 --batch-size 320 --wd 0.04 --wd-end 0.4 --norm-wd 0 --amp --amp-dtype bfloat16 --compile --compile-opt --wds --wds-info /mnt/data/ssl_micro_packed/_info.json",
|
| 3 |
+
"network": "vit_reg4_m16_rms_avg",
|
| 4 |
+
"net_param": null,
|
| 5 |
+
"model_config": null,
|
| 6 |
+
"predictor_embed_dim": 384,
|
| 7 |
+
"predictor_num_heads": 12,
|
| 8 |
+
"predictor_depth": 6,
|
| 9 |
+
"compile": true,
|
| 10 |
+
"compile_opt": true,
|
| 11 |
+
"opt": "adamw",
|
| 12 |
+
"momentum": 0.9,
|
| 13 |
+
"nesterov": false,
|
| 14 |
+
"opt_eps": null,
|
| 15 |
+
"opt_betas": null,
|
| 16 |
+
"opt_alpha": null,
|
| 17 |
+
"lr": 0.001,
|
| 18 |
+
"bias_lr": null,
|
| 19 |
+
"lr_scale": null,
|
| 20 |
+
"lr_scale_type": "linear",
|
| 21 |
+
"wd": 0.04,
|
| 22 |
+
"wd_end": 0.4,
|
| 23 |
+
"norm_wd": 0.0,
|
| 24 |
+
"bias_weight_decay": null,
|
| 25 |
+
"transformer_embedding_decay": null,
|
| 26 |
+
"layer_decay": null,
|
| 27 |
+
"lr_scheduler_update": "epoch",
|
| 28 |
+
"lr_scheduler": "cosine",
|
| 29 |
+
"lr_step_size": 40,
|
| 30 |
+
"lr_steps": null,
|
| 31 |
+
"lr_step_gamma": 0.75,
|
| 32 |
+
"lr_cosine_min": 1e-06,
|
| 33 |
+
"lr_power": 1.0,
|
| 34 |
+
"grad_accum_steps": 1,
|
| 35 |
+
"channels": 3,
|
| 36 |
+
"size": [
|
| 37 |
+
224,
|
| 38 |
+
224
|
| 39 |
+
],
|
| 40 |
+
"batch_size": 320,
|
| 41 |
+
"warmup_epochs": 5,
|
| 42 |
+
"aug_type": "birder",
|
| 43 |
+
"aug_level": 1,
|
| 44 |
+
"use_grayscale": false,
|
| 45 |
+
"ra_num_ops": 2,
|
| 46 |
+
"ra_magnitude": 9,
|
| 47 |
+
"augmix_severity": 3,
|
| 48 |
+
"resize_min_scale": 0.35,
|
| 49 |
+
"re_prob": null,
|
| 50 |
+
"simple_crop": false,
|
| 51 |
+
"rgb_mode": "birder",
|
| 52 |
+
"epochs": 60,
|
| 53 |
+
"stop_epoch": 61,
|
| 54 |
+
"save_frequency": 1,
|
| 55 |
+
"keep_last": null,
|
| 56 |
+
"resume_epoch": null,
|
| 57 |
+
"load_states": false,
|
| 58 |
+
"load_scheduler": false,
|
| 59 |
+
"tag": null,
|
| 60 |
+
"log_interval": 100,
|
| 61 |
+
"num_workers": 16,
|
| 62 |
+
"img_loader": "tv",
|
| 63 |
+
"prefetch_factor": null,
|
| 64 |
+
"model_dtype": "float32",
|
| 65 |
+
"amp": true,
|
| 66 |
+
"amp_dtype": "bfloat16",
|
| 67 |
+
"fast_matmul": false,
|
| 68 |
+
"grad_anomaly_detection": false,
|
| 69 |
+
"world_size": 2,
|
| 70 |
+
"dist_url": "env://",
|
| 71 |
+
"find_unused_parameters": false,
|
| 72 |
+
"clip_grad_norm": null,
|
| 73 |
+
"gpu": 0,
|
| 74 |
+
"cpu": false,
|
| 75 |
+
"use_deterministic_algorithms": false,
|
| 76 |
+
"plot_lr": false,
|
| 77 |
+
"no_summary": false,
|
| 78 |
+
"data_path": [],
|
| 79 |
+
"wds": true,
|
| 80 |
+
"wds_info": "/mnt/data/ssl_micro_packed/_info.json",
|
| 81 |
+
"wds_cache_dir": null,
|
| 82 |
+
"wds_train_size": null,
|
| 83 |
+
"wds_split": "training",
|
| 84 |
+
"rank": 0,
|
| 85 |
+
"distributed": true,
|
| 86 |
+
"dist_backend": "nccl"
|
| 87 |
+
}
|