grodino commited on
Commit
eacf11d
1 Parent(s): a008765

Upload timm_training_args.yaml with huggingface_hub

Browse files
Files changed (1) hide show
  1. timm_training_args.yaml +156 -0
timm_training_args.yaml ADDED
@@ -0,0 +1,156 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ aa: null
2
+ amp: false
3
+ amp_dtype: float16
4
+ amp_impl: native
5
+ aug_repeats: 0
6
+ aug_splits: 0
7
+ batch_size: 256
8
+ bce_loss: false
9
+ bce_pos_weight: null
10
+ bce_sum: false
11
+ bce_target_thresh: null
12
+ bn_eps: null
13
+ bn_momentum: null
14
+ channels_last: false
15
+ checkpoint_hist: 10
16
+ class_map: ''
17
+ clip_grad: null
18
+ clip_mode: norm
19
+ color_jitter: 0.4
20
+ color_jitter_prob: null
21
+ cooldown_epochs: 0
22
+ crop_pct: null
23
+ cutmix: 0.0
24
+ cutmix_minmax: null
25
+ data: null
26
+ data_dir: generated/data
27
+ dataset: torch/cifar10
28
+ dataset_download: true
29
+ decay_epochs: 90
30
+ decay_milestones:
31
+ - 90
32
+ - 180
33
+ - 270
34
+ decay_rate: 0.1
35
+ device: cuda
36
+ device_modules: null
37
+ dist_bn: reduce
38
+ drop: 0.0
39
+ drop_block: null
40
+ drop_connect: null
41
+ drop_path: null
42
+ epoch_repeats: 0.0
43
+ epochs: 200
44
+ eval_metric: top1
45
+ experiment: resnet18_cifar10
46
+ fast_norm: false
47
+ fuser: ''
48
+ gaussian_blur_prob: null
49
+ gp: null
50
+ grad_accum_steps: 1
51
+ grad_checkpointing: false
52
+ grayscale_prob: null
53
+ head_init_bias: null
54
+ head_init_scale: null
55
+ hflip: 0.5
56
+ img_size: null
57
+ in_chans: null
58
+ initial_checkpoint: ''
59
+ input_img_mode: null
60
+ input_key: null
61
+ input_size:
62
+ - 3
63
+ - 224
64
+ - 224
65
+ interpolation: ''
66
+ jsd_loss: false
67
+ layer_decay: null
68
+ local_rank: 0
69
+ log_interval: 50
70
+ log_wandb: false
71
+ lr: 0.1
72
+ lr_base: 0.1
73
+ lr_base_scale: ''
74
+ lr_base_size: 256
75
+ lr_cycle_decay: 0.5
76
+ lr_cycle_limit: 1
77
+ lr_cycle_mul: 1.0
78
+ lr_k_decay: 1.0
79
+ lr_noise: null
80
+ lr_noise_pct: 0.67
81
+ lr_noise_std: 1.0
82
+ mean:
83
+ - 0.4914
84
+ - 0.4822
85
+ - 0.4465
86
+ min_lr: 0
87
+ mixup: 0.0
88
+ mixup_mode: batch
89
+ mixup_off_epoch: 0
90
+ mixup_prob: 1.0
91
+ mixup_switch_prob: 0.5
92
+ model: resnet18
93
+ model_ema: false
94
+ model_ema_decay: 0.9998
95
+ model_ema_force_cpu: false
96
+ model_ema_warmup: false
97
+ model_kwargs: {}
98
+ momentum: 0.9
99
+ no_aug: false
100
+ no_ddp_bb: false
101
+ no_prefetcher: false
102
+ no_resume_opt: false
103
+ num_classes: 10
104
+ opt: nesterov
105
+ opt_betas: null
106
+ opt_eps: null
107
+ opt_kwargs: {}
108
+ output: generated/models
109
+ patience_epochs: 3
110
+ pin_mem: false
111
+ pretrained: false
112
+ pretrained_path: null
113
+ ratio:
114
+ - 0.75
115
+ - 1.3333333333333333
116
+ recount: 1
117
+ recovery_interval: 0
118
+ remode: pixel
119
+ reprob: 0.0
120
+ resplit: false
121
+ resume: ''
122
+ save_images: false
123
+ scale:
124
+ - 0.08
125
+ - 1.0
126
+ sched: plateau
127
+ sched_on_updates: false
128
+ seed: 1
129
+ smoothing: 0.1
130
+ split_bn: false
131
+ start_epoch: null
132
+ std:
133
+ - 0.247
134
+ - 0.243
135
+ - 0.261
136
+ sync_bn: false
137
+ synchronize_step: false
138
+ target_key: null
139
+ torchcompile: null
140
+ torchscript: false
141
+ train_crop_mode: null
142
+ train_interpolation: random
143
+ train_num_samples: null
144
+ train_split: train
145
+ tta: 0
146
+ use_multi_epochs_loader: false
147
+ val_num_samples: null
148
+ val_split: validation
149
+ validation_batch_size: null
150
+ vflip: 0.0
151
+ warmup_epochs: 5
152
+ warmup_lr: 1.0e-05
153
+ warmup_prefix: false
154
+ weight_decay: 0.0005
155
+ worker_seeding: all
156
+ workers: 4