pretraining fmri mae
start: 2026-02-06 22:36:27
cwd: /admin/home/connor/fmri-fm
sha: d20519e5856f827bd427b02eb9f01a851fc41e98, status: has uncommitted changes, branch: dev/clane9
config:
name: patch_size/patch8/pretrain
notes: patch_size ablations patch8 (patch_size=8 mask_patch_size=16)
output_dir: experiments/patch_size/output/patch_size/patch8/pretrain
input_space: flat
patch_size: 8
num_frames: 16
t_patch_size: 4
mask_ratio: 0.9
pred_mask_ratio: null
masking: tube
masking_kwargs: {}
mask_patch_size: 16
model: mae_vit_base
model_kwargs:
  decoding: attn
  pos_embed: sep
  target_norm: null
  pca_norm_nc: 2
  t_pred_stride: 2
  no_decode_pos: true
  mask_drop_scale: false
  pred_edge_pad: 0
  gauss_sigma: null
  class_token: true
  reg_tokens: 0
  no_embed_class: true
  head_init_scale: 0.0
  decoder_depth: 4
  drop_path_rate: 0.0
datasets:
  hcp-train:
    type: wds
    url: /data/fmri-datasets/pretrain/hcpya-all.${input_space}.wds/hcpya-all-${input_space}-{00000..01799}.tar
    clipping: random
    clipping_kwargs:
      oversample: 4.0
    shuffle: true
    buffer_size: 2000
    samples_per_epoch: 200000
  hcp-train-subset:
    type: arrow
    root: s3://medarc/fmri-datasets/eval/hcpya-clips.${input_space}.arrow/validation
    split_range:
    - 0
    - 2000
    shuffle: false
  hcp-val:
    type: arrow
    root: s3://medarc/fmri-datasets/eval/hcpya-clips.${input_space}.arrow/test
    split_range:
    - 0
    - 2000
    shuffle: false
train_dataset: hcp-train
eval_datasets:
- hcp-train-subset
- hcp-val
val_dataset: null
clip_vmax: 3.0
normalize: frame
tr_scale: null
crop_scale: null
crop_aspect: null
gray_jitter: null
num_workers: 16
epochs: 100
batch_size: 32
accum_iter: 1
base_lr: 0.001
min_lr: 0.0
warmup_epochs: 5
weight_decay: 0.05
betas:
- 0.9
- 0.95
clip_grad: 1.0
amp: true
amp_dtype: float16
ckpt: null
resume: true
auto_resume: true
start_epoch: 0
max_checkpoints: 0
checkpoint_period: null
plot_period: 1
device: cuda
presend_cuda: false
seed: 6500
debug: false
wandb: true
wandb_entity: null
wandb_project: fMRI-foundation-model
rank: 0
world_size: 1
gpu: 0
distributed: true
dist_backend: nccl
in_chans: 1
img_size:
- 224
- 560

train transform:
Compose(
      ToTensor()
      TemporalCenterCrop(num_frames=16)
      Normalize(mode='frame')
      Clip(vmax=3.0)
      FlatUnmask((224, 560))
)
val transform:
Compose(
      ToTensor()
      TemporalCenterCrop(num_frames=16)
      Normalize(mode='frame')
      Clip(vmax=3.0)
      FlatUnmask((224, 560))
)
mask generator:
TubeMasking(
  mask_ratio=0.9
  (patchify): Patchify2D((224, 560), (16, 16), in_chans=1)
)
loading dataset: hcp-train

type: wds
url: /data/fmri-datasets/pretrain/hcpya-all.${input_space}.wds/hcpya-all-${input_space}-{00000..01799}.tar
clipping: random
clipping_kwargs:
  oversample: 4.0
shuffle: true
buffer_size: 2000
samples_per_epoch: 200000

loading dataset: hcp-train-subset

type: arrow
root: s3://medarc/fmri-datasets/eval/hcpya-clips.${input_space}.arrow/validation
split_range:
- 0
- 2000
shuffle: false

split indices: [899, 472, 767, 116, 1265, 1852, 300, 1335, 361, 1560]
loading dataset: hcp-val

type: arrow
root: s3://medarc/fmri-datasets/eval/hcpya-clips.${input_space}.arrow/test
split_range:
- 0
- 2000
shuffle: false

split indices: [1075, 1189, 738, 1350, 965, 1964, 1367, 1183, 1619, 1407]
model:
MaskedAutoencoderViT(
  decoding=attn, t_pred_stride=2, pred_edge_pad=0, no_decode_pos=True
  (encoder): MaskedEncoder(
    class_token=True, reg_tokens=0, no_embed_class=True, mask_drop_scale=False
    (patchify): Patchify3D((16, 224, 560), (4, 8, 8), in_chans=1)
    (patch_embed): Linear(in_features=256, out_features=768, bias=True)
    (pos_embed): SeparablePosEmbed(768, (4, 28, 70))
    (blocks): ModuleList(
      (0-11): 12 x Block(
        (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
        (attn): Attention(
          num_heads=12
          (q): Linear(in_features=768, out_features=768, bias=True)
          (k): Linear(in_features=768, out_features=768, bias=True)
          (v): Linear(in_features=768, out_features=768, bias=True)
          (proj): Linear(in_features=768, out_features=768, bias=True)
        )
        (drop_path1): Identity()
        (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
        (mlp): Mlp(
          (fc1): Linear(in_features=768, out_features=3072, bias=True)
          (act): GELU(approximate='none')
          (fc2): Linear(in_features=3072, out_features=768, bias=True)
        )
        (drop_path2): Identity()
      )
    )
    (norm): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
  )
  (pred_patchify): StridedPatchify3D((16, 224, 560), (2, 8, 8), in_chans=1, t_stride=2)
  (decoder): MaskedDecoder(
    cross_decode=False, class_token=True, no_embed_class=True
    (pos_embed): SeparablePosEmbed(512, (4, 28, 70))
    (proj): Linear(in_features=768, out_features=512, bias=True)
    (blocks): ModuleList(
      (0-3): 4 x Block(
        (norm1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
        (attn): Attention(
          num_heads=16
          (q): Linear(in_features=512, out_features=512, bias=True)
          (k): Linear(in_features=512, out_features=512, bias=True)
          (v): Linear(in_features=512, out_features=512, bias=True)
          (proj): Linear(in_features=512, out_features=512, bias=True)
        )
        (drop_path1): Identity()
        (norm2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
        (mlp): Mlp(
          (fc1): Linear(in_features=512, out_features=2048, bias=True)
          (act): GELU(approximate='none')
          (fc2): Linear(in_features=2048, out_features=512, bias=True)
        )
        (drop_path2): Identity()
      )
    )
    (norm): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
    (head): Linear(in_features=512, out_features=128, bias=True)
  )
)
num params: 100.8M
total batch size: 32 = 32 bs per gpu x 1 accum x 1 gpus
lr: 1.25e-04 = 1.00e-03 x 32 / 256
full schedule: epochs = 100 (steps = 625000)
warmup: epochs = 5 (steps = 31250)
start training for 100 epochs
Train: [0]  [   0/6250]  eta: 20:17:30  lr: 0.000000  grad: 0.0257 (0.0257)  loss: 0.9975 (0.9975)  time: 11.6881  data: 9.0837  max mem: 24896
Train: [0]  [ 100/6250]  eta: 0:48:07  lr: 0.000000  grad: 0.0130 (0.0141)  loss: 0.9959 (0.9961)  time: 0.2674  data: 0.0002  max mem: 26028
Train: [0]  [ 200/6250]  eta: 0:37:14  lr: 0.000001  grad: 0.0112 (0.0133)  loss: 0.9963 (0.9959)  time: 0.2680  data: 0.0002  max mem: 26028
Train: [0]  [ 300/6250]  eta: 0:33:18  lr: 0.000001  grad: 0.0110 (0.0127)  loss: 0.9960 (0.9959)  time: 0.2714  data: 0.0002  max mem: 26030
Train: [0]  [ 400/6250]  eta: 0:31:05  lr: 0.000002  grad: 0.0117 (0.0124)  loss: 0.9959 (0.9959)  time: 0.2685  data: 0.0002  max mem: 26030
Train: [0]  [ 500/6250]  eta: 0:29:35  lr: 0.000002  grad: 0.0119 (0.0123)  loss: 0.9957 (0.9959)  time: 0.2665  data: 0.0002  max mem: 26030
Train: [0]  [ 600/6250]  eta: 0:28:25  lr: 0.000002  grad: 0.0116 (0.0121)  loss: 0.9961 (0.9959)  time: 0.2678  data: 0.0002  max mem: 26030
Train: [0]  [ 700/6250]  eta: 0:27:28  lr: 0.000003  grad: 0.0109 (0.0120)  loss: 0.9957 (0.9959)  time: 0.2671  data: 0.0002  max mem: 26030
Train: [0]  [ 800/6250]  eta: 0:26:39  lr: 0.000003  grad: 0.0125 (0.0120)  loss: 0.9956 (0.9959)  time: 0.2674  data: 0.0002  max mem: 26030
Train: [0]  [ 900/6250]  eta: 0:25:53  lr: 0.000004  grad: 0.0117 (0.0119)  loss: 0.9954 (0.9959)  time: 0.2660  data: 0.0001  max mem: 26030
Train: [0]  [1000/6250]  eta: 0:25:12  lr: 0.000004  grad: 0.0122 (0.0119)  loss: 0.9959 (0.9959)  time: 0.2671  data: 0.0002  max mem: 26075
Train: [0]  [1100/6250]  eta: 0:24:34  lr: 0.000004  grad: 0.0139 (0.0120)  loss: 0.9960 (0.9959)  time: 0.2664  data: 0.0002  max mem: 26075
Train: [0]  [1200/6250]  eta: 0:23:57  lr: 0.000005  grad: 0.0167 (0.0124)  loss: 0.9957 (0.9959)  time: 0.2677  data: 0.0002  max mem: 26075
Train: [0]  [1300/6250]  eta: 0:23:22  lr: 0.000005  grad: 0.0200 (0.0130)  loss: 0.9955 (0.9959)  time: 0.2685  data: 0.0002  max mem: 26075
Train: [0]  [1400/6250]  eta: 0:22:49  lr: 0.000006  grad: 0.0236 (0.0138)  loss: 0.9956 (0.9959)  time: 0.2691  data: 0.0002  max mem: 26075
Train: [0]  [1500/6250]  eta: 0:22:16  lr: 0.000006  grad: 0.0300 (0.0153)  loss: 0.9954 (0.9958)  time: 0.2700  data: 0.0002  max mem: 26075
Train: [0]  [1600/6250]  eta: 0:21:44  lr: 0.000006  grad: 0.0367 (0.0171)  loss: 0.9951 (0.9958)  time: 0.2682  data: 0.0002  max mem: 26075
Train: [0]  [1700/6250]  eta: 0:21:13  lr: 0.000007  grad: 0.0554 (0.0192)  loss: 0.9950 (0.9958)  time: 0.2677  data: 0.0002  max mem: 26075
Train: [0]  [1800/6250]  eta: 0:20:41  lr: 0.000007  grad: 0.0417 (0.0212)  loss: 0.9950 (0.9957)  time: 0.2666  data: 0.0001  max mem: 26075
Train: [0]  [1900/6250]  eta: 0:20:11  lr: 0.000008  grad: 0.0409 (0.0227)  loss: 0.9950 (0.9957)  time: 0.2673  data: 0.0002  max mem: 26075
Train: [0]  [2000/6250]  eta: 0:19:41  lr: 0.000008  grad: 0.0420 (0.0244)  loss: 0.9949 (0.9956)  time: 0.2691  data: 0.0002  max mem: 26075
Train: [0]  [2100/6250]  eta: 0:19:11  lr: 0.000008  grad: 0.0411 (0.0254)  loss: 0.9953 (0.9956)  time: 0.2702  data: 0.0002  max mem: 26075
Train: [0]  [2200/6250]  eta: 0:18:41  lr: 0.000009  grad: 0.0537 (0.0266)  loss: 0.9948 (0.9956)  time: 0.2680  data: 0.0002  max mem: 26075
Train: [0]  [2300/6250]  eta: 0:18:12  lr: 0.000009  grad: 0.0495 (0.0279)  loss: 0.9947 (0.9955)  time: 0.2679  data: 0.0002  max mem: 26075
Train: [0]  [2400/6250]  eta: 0:17:43  lr: 0.000010  grad: 0.0446 (0.0289)  loss: 0.9943 (0.9955)  time: 0.2670  data: 0.0002  max mem: 26075
Train: [0]  [2500/6250]  eta: 0:17:14  lr: 0.000010  grad: 0.0471 (0.0298)  loss: 0.9941 (0.9955)  time: 0.2687  data: 0.0002  max mem: 26075
Train: [0]  [2600/6250]  eta: 0:16:46  lr: 0.000010  grad: 0.0435 (0.0306)  loss: 0.9949 (0.9954)  time: 0.2681  data: 0.0002  max mem: 26075
Train: [0]  [2700/6250]  eta: 0:16:17  lr: 0.000011  grad: 0.0483 (0.0314)  loss: 0.9940 (0.9954)  time: 0.2664  data: 0.0002  max mem: 26075
Train: [0]  [2800/6250]  eta: 0:15:49  lr: 0.000011  grad: 0.0494 (0.0323)  loss: 0.9939 (0.9953)  time: 0.2684  data: 0.0002  max mem: 26075
Train: [0]  [2900/6250]  eta: 0:15:20  lr: 0.000012  grad: 0.0605 (0.0333)  loss: 0.9937 (0.9953)  time: 0.2666  data: 0.0002  max mem: 26075
Train: [0]  [3000/6250]  eta: 0:14:52  lr: 0.000012  grad: 0.0527 (0.0342)  loss: 0.9938 (0.9952)  time: 0.2679  data: 0.0002  max mem: 26075
Train: [0]  [3100/6250]  eta: 0:14:24  lr: 0.000012  grad: 0.0684 (0.0351)  loss: 0.9930 (0.9952)  time: 0.2673  data: 0.0002  max mem: 26075
Train: [0]  [3200/6250]  eta: 0:13:56  lr: 0.000013  grad: 0.0683 (0.0362)  loss: 0.9933 (0.9951)  time: 0.2675  data: 0.0002  max mem: 26075
Train: [0]  [3300/6250]  eta: 0:13:28  lr: 0.000013  grad: 0.0632 (0.0373)  loss: 0.9922 (0.9950)  time: 0.2683  data: 0.0002  max mem: 26075
Train: [0]  [3400/6250]  eta: 0:13:00  lr: 0.000014  grad: 0.0781 (0.0384)  loss: 0.9920 (0.9949)  time: 0.2676  data: 0.0001  max mem: 26075
Train: [0]  [3500/6250]  eta: 0:12:32  lr: 0.000014  grad: 0.0624 (0.0394)  loss: 0.9926 (0.9949)  time: 0.2672  data: 0.0002  max mem: 26075
Train: [0]  [3600/6250]  eta: 0:12:04  lr: 0.000014  grad: 0.0685 (0.0405)  loss: 0.9918 (0.9948)  time: 0.2678  data: 0.0002  max mem: 26075
Train: [0]  [3700/6250]  eta: 0:11:36  lr: 0.000015  grad: 0.0722 (0.0417)  loss: 0.9920 (0.9947)  time: 0.2679  data: 0.0002  max mem: 26075
Train: [0]  [3800/6250]  eta: 0:11:09  lr: 0.000015  grad: 0.0724 (0.0427)  loss: 0.9906 (0.9946)  time: 0.2672  data: 0.0002  max mem: 26075
Train: [0]  [3900/6250]  eta: 0:10:41  lr: 0.000016  grad: 0.0691 (0.0436)  loss: 0.9916 (0.9945)  time: 0.2682  data: 0.0001  max mem: 26075
Train: [0]  [4000/6250]  eta: 0:10:14  lr: 0.000016  grad: 0.0709 (0.0444)  loss: 0.9925 (0.9945)  time: 0.2675  data: 0.0002  max mem: 26075
Train: [0]  [4100/6250]  eta: 0:09:46  lr: 0.000016  grad: 0.0766 (0.0453)  loss: 0.9921 (0.9944)  time: 0.2689  data: 0.0002  max mem: 26075
Train: [0]  [4200/6250]  eta: 0:09:18  lr: 0.000017  grad: 0.0719 (0.0461)  loss: 0.9910 (0.9943)  time: 0.2686  data: 0.0002  max mem: 26075
Train: [0]  [4300/6250]  eta: 0:08:51  lr: 0.000017  grad: 0.0859 (0.0470)  loss: 0.9913 (0.9942)  time: 0.2675  data: 0.0002  max mem: 26075
Train: [0]  [4400/6250]  eta: 0:08:24  lr: 0.000018  grad: 0.0688 (0.0477)  loss: 0.9914 (0.9942)  time: 0.2690  data: 0.0002  max mem: 26075
Train: [0]  [4500/6250]  eta: 0:07:56  lr: 0.000018  grad: 0.0776 (0.0485)  loss: 0.9898 (0.9941)  time: 0.2686  data: 0.0002  max mem: 26075
Train: [0]  [4600/6250]  eta: 0:07:29  lr: 0.000018  grad: 0.0893 (0.0492)  loss: 0.9905 (0.9940)  time: 0.2678  data: 0.0002  max mem: 26075
Train: [0]  [4700/6250]  eta: 0:07:01  lr: 0.000019  grad: 0.0777 (0.0500)  loss: 0.9892 (0.9939)  time: 0.2670  data: 0.0002  max mem: 26075
Train: [0]  [4800/6250]  eta: 0:06:34  lr: 0.000019  grad: 0.0714 (0.0506)  loss: 0.9914 (0.9938)  time: 0.2675  data: 0.0002  max mem: 26075
Train: [0]  [4900/6250]  eta: 0:06:07  lr: 0.000020  grad: 0.0625 (0.0511)  loss: 0.9927 (0.9938)  time: 0.2667  data: 0.0001  max mem: 26075
Train: [0]  [5000/6250]  eta: 0:05:39  lr: 0.000020  grad: 0.0864 (0.0517)  loss: 0.9912 (0.9937)  time: 0.2683  data: 0.0002  max mem: 26109
Train: [0]  [5100/6250]  eta: 0:05:12  lr: 0.000020  grad: 0.0661 (0.0522)  loss: 0.9913 (0.9937)  time: 0.2676  data: 0.0002  max mem: 26109
Train: [0]  [5200/6250]  eta: 0:04:45  lr: 0.000021  grad: 0.0631 (0.0527)  loss: 0.9913 (0.9936)  time: 0.2668  data: 0.0001  max mem: 26109
Train: [0]  [5300/6250]  eta: 0:04:18  lr: 0.000021  grad: 0.0752 (0.0533)  loss: 0.9901 (0.9936)  time: 0.2679  data: 0.0001  max mem: 26109
Train: [0]  [5400/6250]  eta: 0:03:50  lr: 0.000022  grad: 0.0788 (0.0537)  loss: 0.9902 (0.9935)  time: 0.2689  data: 0.0002  max mem: 26109
Train: [0]  [5500/6250]  eta: 0:03:23  lr: 0.000022  grad: 0.0828 (0.0542)  loss: 0.9915 (0.9935)  time: 0.2682  data: 0.0002  max mem: 26109
Train: [0]  [5600/6250]  eta: 0:02:56  lr: 0.000022  grad: 0.0630 (0.0546)  loss: 0.9920 (0.9934)  time: 0.2677  data: 0.0002  max mem: 26109
Train: [0]  [5700/6250]  eta: 0:02:29  lr: 0.000023  grad: 0.0730 (0.0549)  loss: 0.9908 (0.9933)  time: 0.2681  data: 0.0002  max mem: 26109
Train: [0]  [5800/6250]  eta: 0:02:02  lr: 0.000023  grad: 0.0736 (0.0554)  loss: 0.9892 (0.9933)  time: 0.2676  data: 0.0002  max mem: 26109
Train: [0]  [5900/6250]  eta: 0:01:34  lr: 0.000024  grad: 0.0697 (0.0560)  loss: 0.9919 (0.9932)  time: 0.2683  data: 0.0002  max mem: 26109
Train: [0]  [6000/6250]  eta: 0:01:07  lr: 0.000024  grad: 0.0845 (0.0564)  loss: 0.9898 (0.9932)  time: 0.2677  data: 0.0002  max mem: 26109
Train: [0]  [6100/6250]  eta: 0:00:40  lr: 0.000024  grad: 0.0712 (0.0569)  loss: 0.9905 (0.9931)  time: 0.2676  data: 0.0002  max mem: 26109
Train: [0]  [6200/6250]  eta: 0:00:13  lr: 0.000025  grad: 0.0710 (0.0574)  loss: 0.9898 (0.9930)  time: 0.2678  data: 0.0002  max mem: 26109
Train: [0]  [6249/6250]  eta: 0:00:00  lr: 0.000025  grad: 0.0812 (0.0575)  loss: 0.9884 (0.9930)  time: 0.2680  data: 0.0002  max mem: 26109
Train: [0] Total time: 0:28:18 (0.2717 s / it)
Averaged stats: lr: 0.000025  grad: 0.0812 (0.0575)  loss: 0.9884 (0.9930)
Eval (hcp-train-subset): [0]  [ 0/62]  eta: 0:04:53  loss: 0.9894 (0.9894)  time: 4.7340  data: 4.6488  max mem: 26109
Eval (hcp-train-subset): [0]  [61/62]  eta: 0:00:00  loss: 0.9917 (0.9908)  time: 0.1183  data: 0.0350  max mem: 26109
Eval (hcp-train-subset): [0] Total time: 0:00:12 (0.1961 s / it)
Averaged stats (hcp-train-subset): loss: 0.9917 (0.9908)
Making plots (hcp-train-subset): example=17
Eval (hcp-val): [0]  [ 0/62]  eta: 0:04:39  loss: 0.9843 (0.9843)  time: 4.5043  data: 4.4184  max mem: 26109
Eval (hcp-val): [0]  [61/62]  eta: 0:00:00  loss: 0.9889 (0.9897)  time: 0.1214  data: 0.0385  max mem: 26109
Eval (hcp-val): [0] Total time: 0:00:11 (0.1925 s / it)
Averaged stats (hcp-val): loss: 0.9889 (0.9897)
Making plots (hcp-val): example=5
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [1]  [   0/6250]  eta: 9:47:54  lr: 0.000025  grad: 0.0858 (0.0858)  loss: 0.9920 (0.9920)  time: 5.6439  data: 5.3693  max mem: 26109
Train: [1]  [ 100/6250]  eta: 0:33:07  lr: 0.000025  grad: 0.0687 (0.0784)  loss: 0.9915 (0.9912)  time: 0.2696  data: 0.0002  max mem: 26109
Train: [1]  [ 200/6250]  eta: 0:29:50  lr: 0.000026  grad: 0.0629 (0.0722)  loss: 0.9916 (0.9914)  time: 0.2682  data: 0.0002  max mem: 26109
Train: [1]  [ 300/6250]  eta: 0:28:29  lr: 0.000026  grad: 0.0776 (0.0729)  loss: 0.9891 (0.9910)  time: 0.2714  data: 0.0002  max mem: 26109
Train: [1]  [ 400/6250]  eta: 0:27:34  lr: 0.000027  grad: 0.0827 (0.0741)  loss: 0.9899 (0.9906)  time: 0.2703  data: 0.0002  max mem: 26109
Train: [1]  [ 500/6250]  eta: 0:26:50  lr: 0.000027  grad: 0.0725 (0.0743)  loss: 0.9905 (0.9905)  time: 0.2702  data: 0.0002  max mem: 26109
Train: [1]  [ 600/6250]  eta: 0:26:11  lr: 0.000027  grad: 0.0626 (0.0748)  loss: 0.9891 (0.9903)  time: 0.2683  data: 0.0002  max mem: 26109
Train: [1]  [ 700/6250]  eta: 0:25:37  lr: 0.000028  grad: 0.0637 (0.0750)  loss: 0.9896 (0.9902)  time: 0.2698  data: 0.0002  max mem: 26109
Train: [1]  [ 800/6250]  eta: 0:25:20  lr: 0.000028  grad: 0.0683 (0.0744)  loss: 0.9908 (0.9902)  time: 0.2734  data: 0.0003  max mem: 26109
Train: [1]  [ 900/6250]  eta: 0:24:52  lr: 0.000029  grad: 0.0760 (0.0748)  loss: 0.9896 (0.9902)  time: 0.2792  data: 0.0002  max mem: 26109
Train: [1]  [1000/6250]  eta: 0:24:22  lr: 0.000029  grad: 0.0699 (0.0748)  loss: 0.9895 (0.9901)  time: 0.2704  data: 0.0002  max mem: 26109
Train: [1]  [1100/6250]  eta: 0:24:08  lr: 0.000029  grad: 0.0803 (0.0748)  loss: 0.9891 (0.9900)  time: 0.4123  data: 0.1366  max mem: 26109
Train: [1]  [1200/6250]  eta: 0:23:42  lr: 0.000030  grad: 0.0729 (0.0755)  loss: 0.9897 (0.9900)  time: 0.2684  data: 0.0002  max mem: 26109
Train: [1]  [1300/6250]  eta: 0:23:09  lr: 0.000030  grad: 0.0721 (0.0757)  loss: 0.9888 (0.9900)  time: 0.2679  data: 0.0002  max mem: 26109
Train: [1]  [1400/6250]  eta: 0:22:36  lr: 0.000031  grad: 0.0777 (0.0762)  loss: 0.9886 (0.9899)  time: 0.2692  data: 0.0002  max mem: 26109
Train: [1]  [1500/6250]  eta: 0:22:05  lr: 0.000031  grad: 0.0769 (0.0767)  loss: 0.9901 (0.9899)  time: 0.2673  data: 0.0002  max mem: 26109
Train: [1]  [1600/6250]  eta: 0:21:34  lr: 0.000031  grad: 0.0756 (0.0771)  loss: 0.9891 (0.9898)  time: 0.2697  data: 0.0002  max mem: 26109
Train: [1]  [1700/6250]  eta: 0:21:04  lr: 0.000032  grad: 0.0692 (0.0776)  loss: 0.9896 (0.9898)  time: 0.2692  data: 0.0002  max mem: 26109
Train: [1]  [1800/6250]  eta: 0:20:39  lr: 0.000032  grad: 0.0666 (0.0778)  loss: 0.9905 (0.9897)  time: 0.2685  data: 0.0002  max mem: 26109
Train: [1]  [1900/6250]  eta: 0:20:17  lr: 0.000033  grad: 0.0683 (0.0779)  loss: 0.9887 (0.9897)  time: 0.2673  data: 0.0002  max mem: 26109
Train: [1]  [2000/6250]  eta: 0:19:47  lr: 0.000033  grad: 0.0934 (0.0782)  loss: 0.9887 (0.9896)  time: 0.2676  data: 0.0002  max mem: 26109
Train: [1]  [2100/6250]  eta: 0:19:17  lr: 0.000033  grad: 0.0738 (0.0783)  loss: 0.9895 (0.9896)  time: 0.2679  data: 0.0002  max mem: 26109
Train: [1]  [2200/6250]  eta: 0:18:47  lr: 0.000034  grad: 0.0770 (0.0784)  loss: 0.9885 (0.9895)  time: 0.2709  data: 0.0002  max mem: 26109
Train: [1]  [2300/6250]  eta: 0:18:18  lr: 0.000034  grad: 0.0802 (0.0787)  loss: 0.9879 (0.9895)  time: 0.2675  data: 0.0002  max mem: 26109
Train: [1]  [2400/6250]  eta: 0:17:48  lr: 0.000035  grad: 0.0745 (0.0790)  loss: 0.9884 (0.9894)  time: 0.2707  data: 0.0002  max mem: 26109
Train: [1]  [2500/6250]  eta: 0:17:19  lr: 0.000035  grad: 0.0855 (0.0789)  loss: 0.9890 (0.9894)  time: 0.2674  data: 0.0002  max mem: 26109
Train: [1]  [2600/6250]  eta: 0:16:50  lr: 0.000035  grad: 0.0722 (0.0790)  loss: 0.9882 (0.9894)  time: 0.2683  data: 0.0002  max mem: 26109
Train: [1]  [2700/6250]  eta: 0:16:21  lr: 0.000036  grad: 0.0756 (0.0790)  loss: 0.9882 (0.9893)  time: 0.2683  data: 0.0001  max mem: 26109
Train: [1]  [2800/6250]  eta: 0:15:53  lr: 0.000036  grad: 0.0785 (0.0794)  loss: 0.9879 (0.9893)  time: 0.2687  data: 0.0002  max mem: 26109
Train: [1]  [2900/6250]  eta: 0:15:24  lr: 0.000037  grad: 0.0788 (0.0794)  loss: 0.9891 (0.9892)  time: 0.2668  data: 0.0002  max mem: 26109
Train: [1]  [3000/6250]  eta: 0:14:56  lr: 0.000037  grad: 0.0705 (0.0796)  loss: 0.9888 (0.9892)  time: 0.2677  data: 0.0002  max mem: 26109
Train: [1]  [3100/6250]  eta: 0:14:27  lr: 0.000037  grad: 0.0760 (0.0797)  loss: 0.9885 (0.9892)  time: 0.2687  data: 0.0002  max mem: 26109
Train: [1]  [3200/6250]  eta: 0:13:59  lr: 0.000038  grad: 0.0861 (0.0799)  loss: 0.9871 (0.9891)  time: 0.2684  data: 0.0002  max mem: 26109
Train: [1]  [3300/6250]  eta: 0:13:31  lr: 0.000038  grad: 0.0871 (0.0801)  loss: 0.9888 (0.9891)  time: 0.2681  data: 0.0002  max mem: 26109
Train: [1]  [3400/6250]  eta: 0:13:03  lr: 0.000039  grad: 0.0676 (0.0801)  loss: 0.9890 (0.9891)  time: 0.2693  data: 0.0002  max mem: 26109
Train: [1]  [3500/6250]  eta: 0:12:35  lr: 0.000039  grad: 0.0769 (0.0800)  loss: 0.9877 (0.9891)  time: 0.2755  data: 0.0030  max mem: 26109
Train: [1]  [3600/6250]  eta: 0:12:08  lr: 0.000039  grad: 0.0756 (0.0801)  loss: 0.9889 (0.9891)  time: 0.2684  data: 0.0001  max mem: 26109
Train: [1]  [3700/6250]  eta: 0:11:40  lr: 0.000040  grad: 0.0814 (0.0801)  loss: 0.9877 (0.9890)  time: 0.2673  data: 0.0002  max mem: 26109
Train: [1]  [3800/6250]  eta: 0:11:12  lr: 0.000040  grad: 0.0643 (0.0800)  loss: 0.9882 (0.9890)  time: 0.2673  data: 0.0002  max mem: 26109
Train: [1]  [3900/6250]  eta: 0:10:44  lr: 0.000041  grad: 0.0639 (0.0799)  loss: 0.9874 (0.9890)  time: 0.2686  data: 0.0002  max mem: 26109
Train: [1]  [4000/6250]  eta: 0:10:16  lr: 0.000041  grad: 0.0721 (0.0798)  loss: 0.9871 (0.9890)  time: 0.2681  data: 0.0002  max mem: 26109
Train: [1]  [4100/6250]  eta: 0:09:48  lr: 0.000041  grad: 0.0671 (0.0797)  loss: 0.9879 (0.9890)  time: 0.2674  data: 0.0002  max mem: 26109
Train: [1]  [4200/6250]  eta: 0:09:21  lr: 0.000042  grad: 0.0593 (0.0795)  loss: 0.9875 (0.9889)  time: 0.2700  data: 0.0002  max mem: 26109
Train: [1]  [4300/6250]  eta: 0:08:53  lr: 0.000042  grad: 0.0657 (0.0794)  loss: 0.9889 (0.9889)  time: 0.2677  data: 0.0002  max mem: 26109
Train: [1]  [4400/6250]  eta: 0:08:26  lr: 0.000043  grad: 0.0696 (0.0794)  loss: 0.9885 (0.9889)  time: 0.2698  data: 0.0002  max mem: 26109
Train: [1]  [4500/6250]  eta: 0:07:58  lr: 0.000043  grad: 0.0738 (0.0793)  loss: 0.9873 (0.9889)  time: 0.2684  data: 0.0002  max mem: 26109
Train: [1]  [4600/6250]  eta: 0:07:31  lr: 0.000043  grad: 0.0784 (0.0792)  loss: 0.9871 (0.9889)  time: 0.2679  data: 0.0002  max mem: 26109
Train: [1]  [4700/6250]  eta: 0:07:03  lr: 0.000044  grad: 0.0800 (0.0792)  loss: 0.9884 (0.9888)  time: 0.2694  data: 0.0002  max mem: 26109
Train: [1]  [4800/6250]  eta: 0:06:36  lr: 0.000044  grad: 0.0776 (0.0791)  loss: 0.9865 (0.9888)  time: 0.2675  data: 0.0002  max mem: 26109
Train: [1]  [4900/6250]  eta: 0:06:08  lr: 0.000045  grad: 0.0820 (0.0791)  loss: 0.9855 (0.9888)  time: 0.2683  data: 0.0002  max mem: 26109
Train: [1]  [5000/6250]  eta: 0:05:41  lr: 0.000045  grad: 0.0759 (0.0792)  loss: 0.9883 (0.9887)  time: 0.2694  data: 0.0002  max mem: 26109
Train: [1]  [5100/6250]  eta: 0:05:13  lr: 0.000045  grad: 0.0682 (0.0792)  loss: 0.9888 (0.9887)  time: 0.2692  data: 0.0002  max mem: 26109
Train: [1]  [5200/6250]  eta: 0:04:46  lr: 0.000046  grad: 0.0786 (0.0791)  loss: 0.9861 (0.9887)  time: 0.2684  data: 0.0002  max mem: 26109
Train: [1]  [5300/6250]  eta: 0:04:19  lr: 0.000046  grad: 0.0719 (0.0792)  loss: 0.9876 (0.9886)  time: 0.2689  data: 0.0002  max mem: 26109
Train: [1]  [5400/6250]  eta: 0:03:51  lr: 0.000047  grad: 0.0711 (0.0793)  loss: 0.9881 (0.9886)  time: 0.2672  data: 0.0002  max mem: 26109
Train: [1]  [5500/6250]  eta: 0:03:24  lr: 0.000047  grad: 0.0809 (0.0793)  loss: 0.9879 (0.9886)  time: 0.2688  data: 0.0001  max mem: 26109
Train: [1]  [5600/6250]  eta: 0:02:57  lr: 0.000047  grad: 0.0767 (0.0794)  loss: 0.9862 (0.9886)  time: 0.2694  data: 0.0002  max mem: 26109
Train: [1]  [5700/6250]  eta: 0:02:29  lr: 0.000048  grad: 0.0791 (0.0794)  loss: 0.9854 (0.9885)  time: 0.2679  data: 0.0002  max mem: 26109
Train: [1]  [5800/6250]  eta: 0:02:02  lr: 0.000048  grad: 0.0706 (0.0794)  loss: 0.9864 (0.9885)  time: 0.2671  data: 0.0002  max mem: 26109
Train: [1]  [5900/6250]  eta: 0:01:35  lr: 0.000049  grad: 0.0791 (0.0794)  loss: 0.9863 (0.9885)  time: 0.2699  data: 0.0002  max mem: 26109
Train: [1]  [6000/6250]  eta: 0:01:08  lr: 0.000049  grad: 0.0807 (0.0794)  loss: 0.9860 (0.9884)  time: 0.2680  data: 0.0002  max mem: 26109
Train: [1]  [6100/6250]  eta: 0:00:40  lr: 0.000049  grad: 0.0833 (0.0795)  loss: 0.9876 (0.9884)  time: 0.2702  data: 0.0002  max mem: 26109
Train: [1]  [6200/6250]  eta: 0:00:13  lr: 0.000050  grad: 0.0808 (0.0795)  loss: 0.9858 (0.9884)  time: 0.2679  data: 0.0002  max mem: 26109
Train: [1]  [6249/6250]  eta: 0:00:00  lr: 0.000050  grad: 0.0756 (0.0795)  loss: 0.9866 (0.9884)  time: 0.2683  data: 0.0002  max mem: 26109
Train: [1] Total time: 0:28:27 (0.2732 s / it)
Averaged stats: lr: 0.000050  grad: 0.0756 (0.0795)  loss: 0.9866 (0.9884)
Eval (hcp-train-subset): [1]  [ 0/62]  eta: 0:03:15  loss: 0.9839 (0.9839)  time: 3.1531  data: 3.0361  max mem: 26109
Eval (hcp-train-subset): [1]  [61/62]  eta: 0:00:00  loss: 0.9901 (0.9893)  time: 0.1396  data: 0.0552  max mem: 26109
Eval (hcp-train-subset): [1] Total time: 0:00:13 (0.2161 s / it)
Averaged stats (hcp-train-subset): loss: 0.9901 (0.9893)
Making plots (hcp-train-subset): example=49
Eval (hcp-val): [1]  [ 0/62]  eta: 0:04:26  loss: 0.9851 (0.9851)  time: 4.2949  data: 4.1707  max mem: 26109
Eval (hcp-val): [1]  [61/62]  eta: 0:00:00  loss: 0.9880 (0.9878)  time: 0.1266  data: 0.0436  max mem: 26109
Eval (hcp-val): [1] Total time: 0:00:12 (0.2055 s / it)
Averaged stats (hcp-val): loss: 0.9880 (0.9878)
Making plots (hcp-val): example=25
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [2]  [   0/6250]  eta: 8:42:34  lr: 0.000050  grad: 0.0762 (0.0762)  loss: 0.9951 (0.9951)  time: 5.0168  data: 4.6043  max mem: 26109
Train: [2]  [ 100/6250]  eta: 0:34:22  lr: 0.000050  grad: 0.0675 (0.0737)  loss: 0.9883 (0.9896)  time: 0.2707  data: 0.0002  max mem: 26109
Train: [2]  [ 200/6250]  eta: 0:30:34  lr: 0.000051  grad: 0.0645 (0.0717)  loss: 0.9889 (0.9892)  time: 0.2715  data: 0.0002  max mem: 26109
Train: [2]  [ 300/6250]  eta: 0:28:59  lr: 0.000051  grad: 0.0769 (0.0729)  loss: 0.9883 (0.9888)  time: 0.2717  data: 0.0002  max mem: 26109
Train: [2]  [ 400/6250]  eta: 0:28:01  lr: 0.000052  grad: 0.0709 (0.0740)  loss: 0.9890 (0.9884)  time: 0.2709  data: 0.0004  max mem: 26109
Train: [2]  [ 500/6250]  eta: 0:27:13  lr: 0.000052  grad: 0.0822 (0.0750)  loss: 0.9859 (0.9880)  time: 0.2699  data: 0.0002  max mem: 26109
Train: [2]  [ 600/6250]  eta: 0:26:31  lr: 0.000052  grad: 0.1074 (0.0773)  loss: 0.9849 (0.9876)  time: 0.2689  data: 0.0002  max mem: 26109
Train: [2]  [ 700/6250]  eta: 0:25:56  lr: 0.000053  grad: 0.0793 (0.0790)  loss: 0.9859 (0.9872)  time: 0.2758  data: 0.0003  max mem: 26109
Train: [2]  [ 800/6250]  eta: 0:25:21  lr: 0.000053  grad: 0.0788 (0.0804)  loss: 0.9856 (0.9870)  time: 0.2718  data: 0.0002  max mem: 26109
Train: [2]  [ 900/6250]  eta: 0:24:48  lr: 0.000054  grad: 0.0881 (0.0809)  loss: 0.9868 (0.9868)  time: 0.2677  data: 0.0002  max mem: 26109
Train: [2]  [1000/6250]  eta: 0:24:16  lr: 0.000054  grad: 0.0777 (0.0818)  loss: 0.9864 (0.9867)  time: 0.2717  data: 0.0002  max mem: 26109
Train: [2]  [1100/6250]  eta: 0:23:45  lr: 0.000054  grad: 0.0960 (0.0830)  loss: 0.9867 (0.9866)  time: 0.2691  data: 0.0002  max mem: 26109
Train: [2]  [1200/6250]  eta: 0:23:14  lr: 0.000055  grad: 0.0778 (0.0836)  loss: 0.9873 (0.9865)  time: 0.2706  data: 0.0002  max mem: 26109
Train: [2]  [1300/6250]  eta: 0:22:44  lr: 0.000055  grad: 0.1060 (0.0845)  loss: 0.9851 (0.9864)  time: 0.2702  data: 0.0002  max mem: 26109
Train: [2]  [1400/6250]  eta: 0:22:15  lr: 0.000056  grad: 0.0728 (0.0846)  loss: 0.9889 (0.9864)  time: 0.2700  data: 0.0002  max mem: 26110
Train: [2]  [1500/6250]  eta: 0:21:45  lr: 0.000056  grad: 0.0744 (0.0850)  loss: 0.9849 (0.9864)  time: 0.2682  data: 0.0002  max mem: 26110
Train: [2]  [1600/6250]  eta: 0:21:16  lr: 0.000056  grad: 0.0823 (0.0852)  loss: 0.9844 (0.9863)  time: 0.2696  data: 0.0002  max mem: 26110
Train: [2]  [1700/6250]  eta: 0:20:48  lr: 0.000057  grad: 0.0902 (0.0857)  loss: 0.9881 (0.9863)  time: 0.2722  data: 0.0002  max mem: 26110
Train: [2]  [1800/6250]  eta: 0:20:20  lr: 0.000057  grad: 0.0846 (0.0859)  loss: 0.9856 (0.9862)  time: 0.2690  data: 0.0002  max mem: 26110
Train: [2]  [1900/6250]  eta: 0:19:51  lr: 0.000058  grad: 0.0843 (0.0863)  loss: 0.9855 (0.9862)  time: 0.2708  data: 0.0002  max mem: 26110
Train: [2]  [2000/6250]  eta: 0:19:23  lr: 0.000058  grad: 0.0833 (0.0863)  loss: 0.9848 (0.9861)  time: 0.2700  data: 0.0002  max mem: 26110
Train: [2]  [2100/6250]  eta: 0:18:55  lr: 0.000058  grad: 0.0891 (0.0868)  loss: 0.9849 (0.9861)  time: 0.2701  data: 0.0002  max mem: 26110
Train: [2]  [2200/6250]  eta: 0:18:27  lr: 0.000059  grad: 0.0771 (0.0871)  loss: 0.9818 (0.9860)  time: 0.2701  data: 0.0002  max mem: 26110
Train: [2]  [2300/6250]  eta: 0:17:59  lr: 0.000059  grad: 0.0817 (0.0874)  loss: 0.9849 (0.9859)  time: 0.2695  data: 0.0002  max mem: 26110
Train: [2]  [2400/6250]  eta: 0:17:31  lr: 0.000060  grad: 0.0947 (0.0885)  loss: 0.9849 (0.9859)  time: 0.2694  data: 0.0002  max mem: 26110
Train: [2]  [2500/6250]  eta: 0:17:03  lr: 0.000060  grad: 0.0938 (0.0889)  loss: 0.9853 (0.9858)  time: 0.2692  data: 0.0002  max mem: 26110
Train: [2]  [2600/6250]  eta: 0:16:35  lr: 0.000060  grad: 0.0862 (0.0902)  loss: 0.9835 (0.9858)  time: 0.2712  data: 0.0002  max mem: 26110
Train: [2]  [2700/6250]  eta: 0:16:08  lr: 0.000061  grad: 0.1211 (0.0920)  loss: 0.9855 (0.9857)  time: 0.2697  data: 0.0002  max mem: 26110
Train: [2]  [2800/6250]  eta: 0:15:40  lr: 0.000061  grad: 0.1551 (0.0943)  loss: 0.9841 (0.9857)  time: 0.2706  data: 0.0002  max mem: 26110
Train: [2]  [2900/6250]  eta: 0:15:13  lr: 0.000062  grad: 0.1230 (0.0970)  loss: 0.9831 (0.9856)  time: 0.2687  data: 0.0002  max mem: 26110
Train: [2]  [3000/6250]  eta: 0:14:45  lr: 0.000062  grad: 0.2307 (0.1017)  loss: 0.9837 (0.9855)  time: 0.2688  data: 0.0002  max mem: 26110
Train: [2]  [3100/6250]  eta: 0:14:18  lr: 0.000062  grad: 0.1516 (0.1055)  loss: 0.9845 (0.9855)  time: 0.2696  data: 0.0002  max mem: 26110
Train: [2]  [3200/6250]  eta: 0:13:50  lr: 0.000063  grad: 0.1336 (0.1099)  loss: 0.9812 (0.9854)  time: 0.2705  data: 0.0002  max mem: 26110
Train: [2]  [3300/6250]  eta: 0:13:23  lr: 0.000063  grad: 0.2516 (0.1161)  loss: 0.9826 (0.9853)  time: 0.2686  data: 0.0002  max mem: 26110
Train: [2]  [3400/6250]  eta: 0:12:55  lr: 0.000064  grad: 0.2072 (0.1214)  loss: 0.9846 (0.9853)  time: 0.2695  data: 0.0001  max mem: 26110
Train: [2]  [3500/6250]  eta: 0:12:28  lr: 0.000064  grad: 0.1772 (0.1265)  loss: 0.9822 (0.9852)  time: 0.2733  data: 0.0002  max mem: 26110
Train: [2]  [3600/6250]  eta: 0:12:01  lr: 0.000064  grad: 0.2627 (0.1318)  loss: 0.9823 (0.9851)  time: 0.2704  data: 0.0002  max mem: 26110
Train: [2]  [3700/6250]  eta: 0:11:33  lr: 0.000065  grad: 0.2471 (0.1371)  loss: 0.9818 (0.9850)  time: 0.2699  data: 0.0002  max mem: 26110
Train: [2]  [3800/6250]  eta: 0:11:06  lr: 0.000065  grad: 0.3020 (0.1428)  loss: 0.9822 (0.9849)  time: 0.2718  data: 0.0002  max mem: 26110
Train: [2]  [3900/6250]  eta: 0:10:39  lr: 0.000066  grad: 0.2026 (0.1463)  loss: 0.9801 (0.9849)  time: 0.2693  data: 0.0002  max mem: 26110
Train: [2]  [4000/6250]  eta: 0:10:11  lr: 0.000066  grad: 0.2315 (0.1504)  loss: 0.9814 (0.9848)  time: 0.2696  data: 0.0002  max mem: 26110
Train: [2]  [4100/6250]  eta: 0:09:44  lr: 0.000066  grad: 0.2624 (0.1539)  loss: 0.9803 (0.9847)  time: 0.2714  data: 0.0002  max mem: 26110
Train: [2]  [4200/6250]  eta: 0:09:17  lr: 0.000067  grad: 0.2620 (0.1564)  loss: 0.9816 (0.9846)  time: 0.2683  data: 0.0002  max mem: 26110
Train: [2]  [4300/6250]  eta: 0:08:50  lr: 0.000067  grad: 0.3585 (0.1593)  loss: 0.9818 (0.9846)  time: 0.2683  data: 0.0002  max mem: 26110
Train: [2]  [4400/6250]  eta: 0:08:23  lr: 0.000068  grad: 0.2511 (0.1629)  loss: 0.9815 (0.9845)  time: 0.2692  data: 0.0002  max mem: 26110
Train: [2]  [4500/6250]  eta: 0:07:55  lr: 0.000068  grad: 0.2963 (0.1656)  loss: 0.9795 (0.9844)  time: 0.2677  data: 0.0002  max mem: 26110
Train: [2]  [4600/6250]  eta: 0:07:28  lr: 0.000068  grad: 0.2889 (0.1678)  loss: 0.9826 (0.9843)  time: 0.2691  data: 0.0002  max mem: 26110
Train: [2]  [4700/6250]  eta: 0:07:01  lr: 0.000069  grad: 0.2283 (0.1694)  loss: 0.9808 (0.9843)  time: 0.2681  data: 0.0002  max mem: 26110
Train: [2]  [4800/6250]  eta: 0:06:33  lr: 0.000069  grad: 0.1988 (0.1717)  loss: 0.9826 (0.9842)  time: 0.2692  data: 0.0002  max mem: 26110
Train: [2]  [4900/6250]  eta: 0:06:06  lr: 0.000070  grad: 0.1871 (0.1729)  loss: 0.9809 (0.9842)  time: 0.2685  data: 0.0002  max mem: 26110
Train: [2]  [5000/6250]  eta: 0:05:39  lr: 0.000070  grad: 0.1754 (0.1753)  loss: 0.9795 (0.9841)  time: 0.2752  data: 0.0002  max mem: 26110
Train: [2]  [5100/6250]  eta: 0:05:12  lr: 0.000070  grad: 0.2308 (0.1774)  loss: 0.9803 (0.9840)  time: 0.2685  data: 0.0002  max mem: 26110
Train: [2]  [5200/6250]  eta: 0:04:45  lr: 0.000071  grad: 0.2696 (0.1797)  loss: 0.9818 (0.9840)  time: 0.2671  data: 0.0002  max mem: 26110
Train: [2]  [5300/6250]  eta: 0:04:17  lr: 0.000071  grad: 0.2332 (0.1818)  loss: 0.9801 (0.9839)  time: 0.2672  data: 0.0001  max mem: 26110
Train: [2]  [5400/6250]  eta: 0:03:50  lr: 0.000072  grad: 0.1745 (0.1835)  loss: 0.9807 (0.9839)  time: 0.2678  data: 0.0002  max mem: 26110
Train: [2]  [5500/6250]  eta: 0:03:23  lr: 0.000072  grad: 0.1721 (0.1845)  loss: 0.9797 (0.9838)  time: 0.2673  data: 0.0002  max mem: 26110
Train: [2]  [5600/6250]  eta: 0:02:56  lr: 0.000072  grad: 0.2637 (0.1859)  loss: 0.9803 (0.9837)  time: 0.2676  data: 0.0002  max mem: 26110
Train: [2]  [5700/6250]  eta: 0:02:29  lr: 0.000073  grad: 0.1561 (0.1886)  loss: 0.9798 (0.9837)  time: 0.2694  data: 0.0002  max mem: 26110
Train: [2]  [5800/6250]  eta: 0:02:02  lr: 0.000073  grad: 0.2680 (0.1895)  loss: 0.9793 (0.9836)  time: 0.2697  data: 0.0002  max mem: 26110
Train: [2]  [5900/6250]  eta: 0:01:34  lr: 0.000074  grad: 0.4017 (0.1910)  loss: 0.9810 (0.9835)  time: 0.2680  data: 0.0002  max mem: 26110
Train: [2]  [6000/6250]  eta: 0:01:07  lr: 0.000074  grad: 0.1476 (0.1918)  loss: 0.9795 (0.9834)  time: 0.2682  data: 0.0002  max mem: 26110
Train: [2]  [6100/6250]  eta: 0:00:40  lr: 0.000074  grad: 0.1772 (0.1927)  loss: 0.9804 (0.9834)  time: 0.2697  data: 0.0002  max mem: 26110
Train: [2]  [6200/6250]  eta: 0:00:13  lr: 0.000075  grad: 0.2078 (0.1941)  loss: 0.9794 (0.9833)  time: 0.2680  data: 0.0002  max mem: 26110
Train: [2]  [6249/6250]  eta: 0:00:00  lr: 0.000075  grad: 0.1658 (0.1945)  loss: 0.9779 (0.9833)  time: 0.2690  data: 0.0002  max mem: 26110
Train: [2] Total time: 0:28:19 (0.2720 s / it)
Averaged stats: lr: 0.000075  grad: 0.1658 (0.1945)  loss: 0.9779 (0.9833)
Eval (hcp-train-subset): [2]  [ 0/62]  eta: 0:04:25  loss: 0.9791 (0.9791)  time: 4.2828  data: 4.1992  max mem: 26110
Eval (hcp-train-subset): [2]  [61/62]  eta: 0:00:00  loss: 0.9816 (0.9799)  time: 0.1280  data: 0.0434  max mem: 26110
Eval (hcp-train-subset): [2] Total time: 0:00:12 (0.2072 s / it)
Averaged stats (hcp-train-subset): loss: 0.9816 (0.9799)
Making plots (hcp-train-subset): example=46
Eval (hcp-val): [2]  [ 0/62]  eta: 0:03:19  loss: 0.9721 (0.9721)  time: 3.2143  data: 3.0906  max mem: 26110
Eval (hcp-val): [2]  [61/62]  eta: 0:00:00  loss: 0.9772 (0.9769)  time: 0.1113  data: 0.0285  max mem: 26110
Eval (hcp-val): [2] Total time: 0:00:12 (0.1997 s / it)
Averaged stats (hcp-val): loss: 0.9772 (0.9769)
Making plots (hcp-val): example=30
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [3]  [   0/6250]  eta: 6:14:22  lr: 0.000075  grad: 0.0815 (0.0815)  loss: 0.9825 (0.9825)  time: 3.5939  data: 3.2303  max mem: 26110
Train: [3]  [ 100/6250]  eta: 0:32:32  lr: 0.000075  grad: 0.2767 (0.2848)  loss: 0.9815 (0.9798)  time: 0.2708  data: 0.0002  max mem: 26110
Train: [3]  [ 200/6250]  eta: 0:29:41  lr: 0.000076  grad: 0.1993 (0.2744)  loss: 0.9803 (0.9791)  time: 0.2693  data: 0.0001  max mem: 26110
Train: [3]  [ 300/6250]  eta: 0:28:24  lr: 0.000076  grad: 0.1408 (0.2740)  loss: 0.9809 (0.9788)  time: 0.2700  data: 0.0002  max mem: 26110
Train: [3]  [ 400/6250]  eta: 0:27:31  lr: 0.000077  grad: 0.3522 (0.2826)  loss: 0.9796 (0.9786)  time: 0.2694  data: 0.0002  max mem: 26110
Train: [3]  [ 500/6250]  eta: 0:26:51  lr: 0.000077  grad: 0.3251 (0.2828)  loss: 0.9806 (0.9784)  time: 0.2717  data: 0.0002  max mem: 26110
Train: [3]  [ 600/6250]  eta: 0:26:15  lr: 0.000077  grad: 0.2116 (0.2876)  loss: 0.9780 (0.9780)  time: 0.2723  data: 0.0003  max mem: 26110
Train: [3]  [ 700/6250]  eta: 0:25:43  lr: 0.000078  grad: 0.2537 (0.2855)  loss: 0.9802 (0.9780)  time: 0.2711  data: 0.0002  max mem: 26110
Train: [3]  [ 800/6250]  eta: 0:25:10  lr: 0.000078  grad: 0.1260 (0.2852)  loss: 0.9766 (0.9777)  time: 0.2743  data: 0.0003  max mem: 26110
Train: [3]  [ 900/6250]  eta: 0:24:39  lr: 0.000079  grad: 0.2361 (0.2809)  loss: 0.9790 (0.9776)  time: 0.2690  data: 0.0001  max mem: 26110
Train: [3]  [1000/6250]  eta: 0:24:10  lr: 0.000079  grad: 0.1771 (0.2764)  loss: 0.9790 (0.9775)  time: 0.2700  data: 0.0002  max mem: 26110
Train: [3]  [1100/6250]  eta: 0:23:40  lr: 0.000079  grad: 0.1845 (0.2733)  loss: 0.9792 (0.9774)  time: 0.2695  data: 0.0002  max mem: 26110
Train: [3]  [1200/6250]  eta: 0:23:10  lr: 0.000080  grad: 0.2347 (0.2750)  loss: 0.9760 (0.9772)  time: 0.2718  data: 0.0002  max mem: 26110
Train: [3]  [1300/6250]  eta: 0:22:41  lr: 0.000080  grad: 0.3489 (0.2761)  loss: 0.9781 (0.9771)  time: 0.2693  data: 0.0002  max mem: 26110
Train: [3]  [1400/6250]  eta: 0:22:12  lr: 0.000081  grad: 0.2259 (0.2768)  loss: 0.9760 (0.9770)  time: 0.2684  data: 0.0002  max mem: 26110
Train: [3]  [1500/6250]  eta: 0:21:44  lr: 0.000081  grad: 0.2334 (0.2746)  loss: 0.9747 (0.9769)  time: 0.2778  data: 0.0002  max mem: 26110
Train: [3]  [1600/6250]  eta: 0:21:16  lr: 0.000081  grad: 0.2666 (0.2750)  loss: 0.9765 (0.9768)  time: 0.2731  data: 0.0002  max mem: 26110
Train: [3]  [1700/6250]  eta: 0:20:47  lr: 0.000082  grad: 0.2504 (0.2762)  loss: 0.9775 (0.9768)  time: 0.2695  data: 0.0002  max mem: 26110
Train: [3]  [1800/6250]  eta: 0:20:19  lr: 0.000082  grad: 0.3191 (0.2762)  loss: 0.9790 (0.9767)  time: 0.2696  data: 0.0002  max mem: 26110
Train: [3]  [1900/6250]  eta: 0:19:51  lr: 0.000083  grad: 0.3110 (0.2758)  loss: 0.9754 (0.9766)  time: 0.2699  data: 0.0002  max mem: 26110
Train: [3]  [2000/6250]  eta: 0:19:23  lr: 0.000083  grad: 0.1918 (0.2734)  loss: 0.9781 (0.9766)  time: 0.2690  data: 0.0002  max mem: 26110
Train: [3]  [2100/6250]  eta: 0:18:54  lr: 0.000083  grad: 0.1766 (0.2736)  loss: 0.9781 (0.9765)  time: 0.2703  data: 0.0002  max mem: 26110
Train: [3]  [2200/6250]  eta: 0:18:26  lr: 0.000084  grad: 0.1644 (0.2721)  loss: 0.9742 (0.9764)  time: 0.2689  data: 0.0002  max mem: 26110
Train: [3]  [2300/6250]  eta: 0:17:59  lr: 0.000084  grad: 0.2724 (0.2721)  loss: 0.9748 (0.9763)  time: 0.2698  data: 0.0002  max mem: 26110
Train: [3]  [2400/6250]  eta: 0:17:31  lr: 0.000085  grad: 0.1686 (0.2715)  loss: 0.9743 (0.9762)  time: 0.2690  data: 0.0002  max mem: 26110
Train: [3]  [2500/6250]  eta: 0:17:03  lr: 0.000085  grad: 0.1825 (0.2720)  loss: 0.9717 (0.9761)  time: 0.2706  data: 0.0002  max mem: 26110
Train: [3]  [2600/6250]  eta: 0:16:35  lr: 0.000085  grad: 0.1430 (0.2715)  loss: 0.9712 (0.9760)  time: 0.2717  data: 0.0002  max mem: 26110
Train: [3]  [2700/6250]  eta: 0:16:08  lr: 0.000086  grad: 0.2504 (0.2706)  loss: 0.9730 (0.9758)  time: 0.2692  data: 0.0001  max mem: 26110
Train: [3]  [2800/6250]  eta: 0:15:40  lr: 0.000086  grad: 0.2433 (0.2715)  loss: 0.9716 (0.9757)  time: 0.2691  data: 0.0002  max mem: 26110
Train: [3]  [2900/6250]  eta: 0:15:13  lr: 0.000087  grad: 0.2025 (0.2720)  loss: 0.9710 (0.9755)  time: 0.2702  data: 0.0001  max mem: 26110
Train: [3]  [3000/6250]  eta: 0:14:45  lr: 0.000087  grad: 0.1586 (0.2711)  loss: 0.9716 (0.9754)  time: 0.2693  data: 0.0002  max mem: 26110
Train: [3]  [3100/6250]  eta: 0:14:18  lr: 0.000087  grad: 0.1873 (0.2695)  loss: 0.9708 (0.9753)  time: 0.2698  data: 0.0002  max mem: 26110
Train: [3]  [3200/6250]  eta: 0:13:50  lr: 0.000088  grad: 0.2377 (0.2701)  loss: 0.9688 (0.9751)  time: 0.2695  data: 0.0002  max mem: 26110
Train: [3]  [3300/6250]  eta: 0:13:23  lr: 0.000088  grad: 0.2537 (0.2695)  loss: 0.9699 (0.9749)  time: 0.2715  data: 0.0002  max mem: 26110
Train: [3]  [3400/6250]  eta: 0:12:55  lr: 0.000089  grad: 0.1920 (0.2688)  loss: 0.9683 (0.9748)  time: 0.2683  data: 0.0001  max mem: 26110
Train: [3]  [3500/6250]  eta: 0:12:28  lr: 0.000089  grad: 0.3256 (0.2686)  loss: 0.9705 (0.9746)  time: 0.2692  data: 0.0001  max mem: 26110
Train: [3]  [3600/6250]  eta: 0:12:01  lr: 0.000089  grad: 0.1778 (0.2695)  loss: 0.9684 (0.9745)  time: 0.2731  data: 0.0002  max mem: 26110
Train: [3]  [3700/6250]  eta: 0:11:33  lr: 0.000090  grad: 0.1577 (0.2689)  loss: 0.9691 (0.9744)  time: 0.2696  data: 0.0002  max mem: 26110
Train: [3]  [3800/6250]  eta: 0:11:06  lr: 0.000090  grad: 0.1752 (0.2683)  loss: 0.9666 (0.9742)  time: 0.2701  data: 0.0002  max mem: 26110
Train: [3]  [3900/6250]  eta: 0:10:39  lr: 0.000091  grad: 0.2748 (0.2686)  loss: 0.9665 (0.9740)  time: 0.2731  data: 0.0002  max mem: 26110
Train: [3]  [4000/6250]  eta: 0:10:11  lr: 0.000091  grad: 0.1670 (0.2688)  loss: 0.9644 (0.9738)  time: 0.2703  data: 0.0002  max mem: 26110
Train: [3]  [4100/6250]  eta: 0:09:44  lr: 0.000091  grad: 0.2044 (0.2679)  loss: 0.9643 (0.9735)  time: 0.2720  data: 0.0002  max mem: 26110
Train: [3]  [4200/6250]  eta: 0:09:18  lr: 0.000092  grad: 0.2079 (0.2677)  loss: 0.9610 (0.9732)  time: 0.2698  data: 0.0002  max mem: 26110
Train: [3]  [4300/6250]  eta: 0:08:50  lr: 0.000092  grad: 0.2563 (0.2686)  loss: 0.9620 (0.9729)  time: 0.2689  data: 0.0002  max mem: 26110
Train: [3]  [4400/6250]  eta: 0:08:23  lr: 0.000093  grad: 0.3087 (0.2689)  loss: 0.9609 (0.9726)  time: 0.2687  data: 0.0002  max mem: 26110
Train: [3]  [4500/6250]  eta: 0:07:56  lr: 0.000093  grad: 0.3407 (0.2695)  loss: 0.9617 (0.9723)  time: 0.2695  data: 0.0002  max mem: 26110
Train: [3]  [4600/6250]  eta: 0:07:28  lr: 0.000093  grad: 0.3007 (0.2706)  loss: 0.9592 (0.9720)  time: 0.2690  data: 0.0002  max mem: 26110
Train: [3]  [4700/6250]  eta: 0:07:01  lr: 0.000094  grad: 0.2101 (0.2711)  loss: 0.9576 (0.9717)  time: 0.2693  data: 0.0002  max mem: 26110
Train: [3]  [4800/6250]  eta: 0:06:34  lr: 0.000094  grad: 0.3212 (0.2719)  loss: 0.9553 (0.9714)  time: 0.2726  data: 0.0002  max mem: 26110
Train: [3]  [4900/6250]  eta: 0:06:07  lr: 0.000095  grad: 0.3142 (0.2723)  loss: 0.9526 (0.9710)  time: 0.2694  data: 0.0002  max mem: 26110
Train: [3]  [5000/6250]  eta: 0:05:39  lr: 0.000095  grad: 0.3365 (0.2734)  loss: 0.9532 (0.9706)  time: 0.2700  data: 0.0002  max mem: 26110
Train: [3]  [5100/6250]  eta: 0:05:12  lr: 0.000095  grad: 0.2748 (0.2738)  loss: 0.9517 (0.9703)  time: 0.2692  data: 0.0002  max mem: 26110
Train: [3]  [5200/6250]  eta: 0:04:45  lr: 0.000096  grad: 0.2685 (0.2743)  loss: 0.9497 (0.9699)  time: 0.2696  data: 0.0001  max mem: 26110
Train: [3]  [5300/6250]  eta: 0:04:18  lr: 0.000096  grad: 0.3572 (0.2753)  loss: 0.9533 (0.9695)  time: 0.2712  data: 0.0002  max mem: 26110
Train: [3]  [5400/6250]  eta: 0:03:51  lr: 0.000097  grad: 0.3000 (0.2760)  loss: 0.9496 (0.9691)  time: 0.2705  data: 0.0002  max mem: 26110
Train: [3]  [5500/6250]  eta: 0:03:23  lr: 0.000097  grad: 0.2749 (0.2762)  loss: 0.9466 (0.9687)  time: 0.2702  data: 0.0001  max mem: 26110
Train: [3]  [5600/6250]  eta: 0:02:56  lr: 0.000097  grad: 0.3545 (0.2769)  loss: 0.9480 (0.9683)  time: 0.2693  data: 0.0002  max mem: 26110
Train: [3]  [5700/6250]  eta: 0:02:29  lr: 0.000098  grad: 0.3418 (0.2783)  loss: 0.9427 (0.9678)  time: 0.2699  data: 0.0001  max mem: 26110
Train: [3]  [5800/6250]  eta: 0:02:02  lr: 0.000098  grad: 0.3611 (0.2794)  loss: 0.9413 (0.9674)  time: 0.2688  data: 0.0002  max mem: 26110
Train: [3]  [5900/6250]  eta: 0:01:35  lr: 0.000099  grad: 0.2490 (0.2800)  loss: 0.9374 (0.9669)  time: 0.2697  data: 0.0002  max mem: 26110
Train: [3]  [6000/6250]  eta: 0:01:08  lr: 0.000099  grad: 0.3172 (0.2808)  loss: 0.9368 (0.9665)  time: 0.2694  data: 0.0002  max mem: 26110
Train: [3]  [6100/6250]  eta: 0:00:40  lr: 0.000099  grad: 0.3907 (0.2818)  loss: 0.9413 (0.9660)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [3]  [6200/6250]  eta: 0:00:13  lr: 0.000100  grad: 0.3175 (0.2834)  loss: 0.9368 (0.9655)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [3]  [6249/6250]  eta: 0:00:00  lr: 0.000100  grad: 0.3701 (0.2840)  loss: 0.9369 (0.9653)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [3] Total time: 0:28:29 (0.2735 s / it)
Averaged stats: lr: 0.000100  grad: 0.3701 (0.2840)  loss: 0.9369 (0.9653)
Eval (hcp-train-subset): [3]  [ 0/62]  eta: 0:03:33  loss: 0.9426 (0.9426)  time: 3.4422  data: 3.3165  max mem: 26157
Eval (hcp-train-subset): [3]  [61/62]  eta: 0:00:00  loss: 0.9401 (0.9384)  time: 0.1138  data: 0.0290  max mem: 26157
Eval (hcp-train-subset): [3] Total time: 0:00:12 (0.2022 s / it)
Averaged stats (hcp-train-subset): loss: 0.9401 (0.9384)
Making plots (hcp-train-subset): example=52
Eval (hcp-val): [3]  [ 0/62]  eta: 0:03:02  loss: 0.9323 (0.9323)  time: 2.9358  data: 2.8099  max mem: 26157
Eval (hcp-val): [3]  [61/62]  eta: 0:00:00  loss: 0.9349 (0.9344)  time: 0.1198  data: 0.0367  max mem: 26157
Eval (hcp-val): [3] Total time: 0:00:12 (0.2093 s / it)
Averaged stats (hcp-val): loss: 0.9349 (0.9344)
Making plots (hcp-val): example=33
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [4]  [   0/6250]  eta: 10:54:53  lr: 0.000100  grad: 0.4930 (0.4930)  loss: 0.9342 (0.9342)  time: 6.2870  data: 6.0115  max mem: 26157
Train: [4]  [ 100/6250]  eta: 0:35:02  lr: 0.000100  grad: 0.3176 (0.3722)  loss: 0.9426 (0.9375)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [4]  [ 200/6250]  eta: 0:30:53  lr: 0.000101  grad: 0.2440 (0.3436)  loss: 0.9386 (0.9367)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [4]  [ 300/6250]  eta: 0:29:11  lr: 0.000101  grad: 0.3442 (0.3450)  loss: 0.9342 (0.9359)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [4]  [ 400/6250]  eta: 0:28:06  lr: 0.000102  grad: 0.4009 (0.3433)  loss: 0.9298 (0.9349)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [4]  [ 500/6250]  eta: 0:27:18  lr: 0.000102  grad: 0.3295 (0.3430)  loss: 0.9322 (0.9343)  time: 0.2736  data: 0.0002  max mem: 26157
Train: [4]  [ 600/6250]  eta: 0:26:37  lr: 0.000102  grad: 0.3349 (0.3445)  loss: 0.9285 (0.9337)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [4]  [ 700/6250]  eta: 0:26:00  lr: 0.000103  grad: 0.2766 (0.3442)  loss: 0.9270 (0.9331)  time: 0.2741  data: 0.0003  max mem: 26157
Train: [4]  [ 800/6250]  eta: 0:25:25  lr: 0.000103  grad: 0.2476 (0.3428)  loss: 0.9313 (0.9328)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [4]  [ 900/6250]  eta: 0:24:52  lr: 0.000104  grad: 0.3016 (0.3415)  loss: 0.9273 (0.9323)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [4]  [1000/6250]  eta: 0:24:20  lr: 0.000104  grad: 0.2983 (0.3404)  loss: 0.9276 (0.9318)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [4]  [1100/6250]  eta: 0:23:48  lr: 0.000104  grad: 0.3702 (0.3421)  loss: 0.9272 (0.9314)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [4]  [1200/6250]  eta: 0:23:17  lr: 0.000105  grad: 0.3041 (0.3437)  loss: 0.9290 (0.9311)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [4]  [1300/6250]  eta: 0:22:47  lr: 0.000105  grad: 0.2478 (0.3406)  loss: 0.9272 (0.9306)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [4]  [1400/6250]  eta: 0:22:17  lr: 0.000106  grad: 0.3095 (0.3384)  loss: 0.9250 (0.9303)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [4]  [1500/6250]  eta: 0:21:48  lr: 0.000106  grad: 0.2952 (0.3390)  loss: 0.9237 (0.9299)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [4]  [1600/6250]  eta: 0:21:19  lr: 0.000106  grad: 0.3573 (0.3388)  loss: 0.9236 (0.9296)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [4]  [1700/6250]  eta: 0:20:50  lr: 0.000107  grad: 0.3191 (0.3393)  loss: 0.9238 (0.9293)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [4]  [1800/6250]  eta: 0:20:22  lr: 0.000107  grad: 0.2943 (0.3393)  loss: 0.9267 (0.9291)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [4]  [1900/6250]  eta: 0:19:54  lr: 0.000108  grad: 0.2244 (0.3371)  loss: 0.9227 (0.9289)  time: 0.2749  data: 0.0002  max mem: 26157
Train: [4]  [2000/6250]  eta: 0:19:26  lr: 0.000108  grad: 0.2634 (0.3360)  loss: 0.9223 (0.9286)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [4]  [2100/6250]  eta: 0:18:58  lr: 0.000108  grad: 0.2574 (0.3362)  loss: 0.9236 (0.9283)  time: 0.2702  data: 0.0001  max mem: 26157
Train: [4]  [2200/6250]  eta: 0:18:30  lr: 0.000109  grad: 0.3094 (0.3353)  loss: 0.9264 (0.9280)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [4]  [2300/6250]  eta: 0:18:02  lr: 0.000109  grad: 0.2654 (0.3356)  loss: 0.9210 (0.9278)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [4]  [2400/6250]  eta: 0:17:37  lr: 0.000110  grad: 0.2588 (0.3350)  loss: 0.9196 (0.9276)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [4]  [2500/6250]  eta: 0:17:09  lr: 0.000110  grad: 0.2461 (0.3341)  loss: 0.9192 (0.9273)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [4]  [2600/6250]  eta: 0:16:41  lr: 0.000110  grad: 0.3244 (0.3345)  loss: 0.9230 (0.9271)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [4]  [2700/6250]  eta: 0:16:13  lr: 0.000111  grad: 0.3775 (0.3349)  loss: 0.9234 (0.9268)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [4]  [2800/6250]  eta: 0:15:44  lr: 0.000111  grad: 0.3044 (0.3339)  loss: 0.9222 (0.9266)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [4]  [2900/6250]  eta: 0:15:16  lr: 0.000112  grad: 0.3170 (0.3330)  loss: 0.9192 (0.9264)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [4]  [3000/6250]  eta: 0:14:48  lr: 0.000112  grad: 0.2305 (0.3328)  loss: 0.9179 (0.9262)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [4]  [3100/6250]  eta: 0:14:21  lr: 0.000112  grad: 0.2680 (0.3319)  loss: 0.9173 (0.9259)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [4]  [3200/6250]  eta: 0:13:53  lr: 0.000113  grad: 0.2466 (0.3304)  loss: 0.9164 (0.9257)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [4]  [3300/6250]  eta: 0:13:25  lr: 0.000113  grad: 0.2484 (0.3294)  loss: 0.9184 (0.9255)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [4]  [3400/6250]  eta: 0:12:58  lr: 0.000114  grad: 0.2573 (0.3286)  loss: 0.9176 (0.9253)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [4]  [3500/6250]  eta: 0:12:30  lr: 0.000114  grad: 0.2669 (0.3277)  loss: 0.9156 (0.9250)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [4]  [3600/6250]  eta: 0:12:03  lr: 0.000114  grad: 0.2784 (0.3273)  loss: 0.9150 (0.9248)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [4]  [3700/6250]  eta: 0:11:35  lr: 0.000115  grad: 0.3003 (0.3271)  loss: 0.9114 (0.9246)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [4]  [3800/6250]  eta: 0:11:07  lr: 0.000115  grad: 0.3616 (0.3266)  loss: 0.9188 (0.9244)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [4]  [3900/6250]  eta: 0:10:40  lr: 0.000116  grad: 0.3130 (0.3256)  loss: 0.9163 (0.9241)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [4]  [4000/6250]  eta: 0:10:13  lr: 0.000116  grad: 0.2573 (0.3245)  loss: 0.9138 (0.9239)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [4]  [4100/6250]  eta: 0:09:46  lr: 0.000116  grad: 0.2459 (0.3237)  loss: 0.9140 (0.9236)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [4]  [4200/6250]  eta: 0:09:18  lr: 0.000117  grad: 0.2672 (0.3231)  loss: 0.9136 (0.9234)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [4]  [4300/6250]  eta: 0:08:51  lr: 0.000117  grad: 0.2435 (0.3223)  loss: 0.9166 (0.9232)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [4]  [4400/6250]  eta: 0:08:24  lr: 0.000118  grad: 0.2322 (0.3223)  loss: 0.9143 (0.9231)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [4]  [4500/6250]  eta: 0:07:56  lr: 0.000118  grad: 0.2934 (0.3223)  loss: 0.9132 (0.9229)  time: 0.2691  data: 0.0003  max mem: 26157
Train: [4]  [4600/6250]  eta: 0:07:29  lr: 0.000118  grad: 0.2788 (0.3213)  loss: 0.9151 (0.9227)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [4]  [4700/6250]  eta: 0:07:01  lr: 0.000119  grad: 0.3466 (0.3207)  loss: 0.9176 (0.9225)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [4]  [4800/6250]  eta: 0:06:34  lr: 0.000119  grad: 0.2863 (0.3197)  loss: 0.9109 (0.9223)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [4]  [4900/6250]  eta: 0:06:07  lr: 0.000120  grad: 0.2392 (0.3195)  loss: 0.9145 (0.9221)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [4]  [5000/6250]  eta: 0:05:40  lr: 0.000120  grad: 0.3337 (0.3196)  loss: 0.9181 (0.9220)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [4]  [5100/6250]  eta: 0:05:12  lr: 0.000120  grad: 0.2476 (0.3186)  loss: 0.9132 (0.9218)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [4]  [5200/6250]  eta: 0:04:45  lr: 0.000121  grad: 0.3021 (0.3181)  loss: 0.9133 (0.9216)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [4]  [5300/6250]  eta: 0:04:18  lr: 0.000121  grad: 0.2428 (0.3172)  loss: 0.9104 (0.9214)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [4]  [5400/6250]  eta: 0:03:51  lr: 0.000122  grad: 0.2788 (0.3168)  loss: 0.9098 (0.9213)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [4]  [5500/6250]  eta: 0:03:23  lr: 0.000122  grad: 0.3624 (0.3160)  loss: 0.9127 (0.9211)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [4]  [5600/6250]  eta: 0:02:56  lr: 0.000122  grad: 0.2094 (0.3154)  loss: 0.9116 (0.9209)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [4]  [5700/6250]  eta: 0:02:29  lr: 0.000123  grad: 0.2660 (0.3144)  loss: 0.9104 (0.9207)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [4]  [5800/6250]  eta: 0:02:02  lr: 0.000123  grad: 0.2497 (0.3136)  loss: 0.9107 (0.9205)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [4]  [5900/6250]  eta: 0:01:35  lr: 0.000124  grad: 0.2111 (0.3131)  loss: 0.9084 (0.9204)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [4]  [6000/6250]  eta: 0:01:08  lr: 0.000124  grad: 0.2697 (0.3123)  loss: 0.9097 (0.9203)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [4]  [6100/6250]  eta: 0:00:40  lr: 0.000124  grad: 0.2316 (0.3113)  loss: 0.9118 (0.9201)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [4]  [6200/6250]  eta: 0:00:13  lr: 0.000125  grad: 0.2301 (0.3103)  loss: 0.9098 (0.9199)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [4]  [6249/6250]  eta: 0:00:00  lr: 0.000125  grad: 0.2027 (0.3098)  loss: 0.9123 (0.9199)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [4] Total time: 0:28:25 (0.2729 s / it)
Averaged stats: lr: 0.000125  grad: 0.2027 (0.3098)  loss: 0.9123 (0.9199)
Eval (hcp-train-subset): [4]  [ 0/62]  eta: 0:04:36  loss: 0.9200 (0.9200)  time: 4.4615  data: 4.3773  max mem: 26157
Eval (hcp-train-subset): [4]  [61/62]  eta: 0:00:00  loss: 0.9149 (0.9142)  time: 0.1307  data: 0.0479  max mem: 26157
Eval (hcp-train-subset): [4] Total time: 0:00:13 (0.2102 s / it)
Averaged stats (hcp-train-subset): loss: 0.9149 (0.9142)
Making plots (hcp-train-subset): example=5
Eval (hcp-val): [4]  [ 0/62]  eta: 0:04:56  loss: 0.9055 (0.9055)  time: 4.7811  data: 4.6969  max mem: 26157
Eval (hcp-val): [4]  [61/62]  eta: 0:00:00  loss: 0.9096 (0.9095)  time: 0.1395  data: 0.0569  max mem: 26157
Eval (hcp-val): [4] Total time: 0:00:13 (0.2138 s / it)
Averaged stats (hcp-val): loss: 0.9096 (0.9095)
Making plots (hcp-val): example=26
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [5]  [   0/6250]  eta: 7:46:06  lr: 0.000125  grad: 0.2023 (0.2023)  loss: 0.9408 (0.9408)  time: 4.4747  data: 4.1715  max mem: 26157
Train: [5]  [ 100/6250]  eta: 0:33:37  lr: 0.000125  grad: 0.1894 (0.2282)  loss: 0.9209 (0.9188)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [5]  [ 200/6250]  eta: 0:30:09  lr: 0.000125  grad: 0.2084 (0.2424)  loss: 0.9167 (0.9158)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [5]  [ 300/6250]  eta: 0:28:40  lr: 0.000125  grad: 0.2713 (0.2478)  loss: 0.9078 (0.9142)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [5]  [ 400/6250]  eta: 0:27:42  lr: 0.000125  grad: 0.2210 (0.2482)  loss: 0.9047 (0.9133)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [5]  [ 500/6250]  eta: 0:26:57  lr: 0.000125  grad: 0.2520 (0.2517)  loss: 0.9104 (0.9124)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [5]  [ 600/6250]  eta: 0:26:20  lr: 0.000125  grad: 0.2976 (0.2526)  loss: 0.9021 (0.9115)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [5]  [ 700/6250]  eta: 0:25:46  lr: 0.000125  grad: 0.2477 (0.2544)  loss: 0.9034 (0.9105)  time: 0.2732  data: 0.0003  max mem: 26157
Train: [5]  [ 800/6250]  eta: 0:25:12  lr: 0.000125  grad: 0.2779 (0.2527)  loss: 0.9069 (0.9101)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [5]  [ 900/6250]  eta: 0:24:39  lr: 0.000125  grad: 0.1818 (0.2491)  loss: 0.9071 (0.9098)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [5]  [1000/6250]  eta: 0:24:07  lr: 0.000125  grad: 0.2165 (0.2474)  loss: 0.9075 (0.9096)  time: 0.2687  data: 0.0003  max mem: 26157
Train: [5]  [1100/6250]  eta: 0:23:36  lr: 0.000125  grad: 0.2313 (0.2441)  loss: 0.9088 (0.9094)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [5]  [1200/6250]  eta: 0:23:06  lr: 0.000125  grad: 0.2476 (0.2446)  loss: 0.9117 (0.9094)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [5]  [1300/6250]  eta: 0:22:37  lr: 0.000125  grad: 0.1916 (0.2425)  loss: 0.9068 (0.9091)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [5]  [1400/6250]  eta: 0:22:07  lr: 0.000125  grad: 0.1607 (0.2413)  loss: 0.9046 (0.9088)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [5]  [1500/6250]  eta: 0:21:38  lr: 0.000125  grad: 0.2400 (0.2428)  loss: 0.9036 (0.9085)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [5]  [1600/6250]  eta: 0:21:09  lr: 0.000125  grad: 0.2060 (0.2416)  loss: 0.9043 (0.9083)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [5]  [1700/6250]  eta: 0:20:41  lr: 0.000125  grad: 0.1847 (0.2410)  loss: 0.9022 (0.9080)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [5]  [1800/6250]  eta: 0:20:13  lr: 0.000125  grad: 0.1960 (0.2412)  loss: 0.9027 (0.9079)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [5]  [1900/6250]  eta: 0:19:45  lr: 0.000125  grad: 0.2372 (0.2410)  loss: 0.9019 (0.9076)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [5]  [2000/6250]  eta: 0:19:17  lr: 0.000125  grad: 0.2391 (0.2402)  loss: 0.9021 (0.9074)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [5]  [2100/6250]  eta: 0:18:50  lr: 0.000125  grad: 0.2601 (0.2400)  loss: 0.9038 (0.9072)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [5]  [2200/6250]  eta: 0:18:22  lr: 0.000125  grad: 0.1872 (0.2391)  loss: 0.9054 (0.9070)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [5]  [2300/6250]  eta: 0:17:54  lr: 0.000125  grad: 0.2056 (0.2388)  loss: 0.9062 (0.9069)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [5]  [2400/6250]  eta: 0:17:27  lr: 0.000125  grad: 0.1923 (0.2381)  loss: 0.9036 (0.9067)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [5]  [2500/6250]  eta: 0:16:59  lr: 0.000125  grad: 0.2455 (0.2381)  loss: 0.8997 (0.9065)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [5]  [2600/6250]  eta: 0:16:32  lr: 0.000125  grad: 0.2332 (0.2380)  loss: 0.9033 (0.9064)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [5]  [2700/6250]  eta: 0:16:05  lr: 0.000125  grad: 0.2179 (0.2373)  loss: 0.9005 (0.9063)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [5]  [2800/6250]  eta: 0:15:38  lr: 0.000125  grad: 0.1955 (0.2365)  loss: 0.9079 (0.9062)  time: 0.2753  data: 0.0002  max mem: 26157
Train: [5]  [2900/6250]  eta: 0:15:10  lr: 0.000125  grad: 0.1696 (0.2359)  loss: 0.9016 (0.9061)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [5]  [3000/6250]  eta: 0:14:43  lr: 0.000125  grad: 0.2088 (0.2347)  loss: 0.9029 (0.9060)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [5]  [3100/6250]  eta: 0:14:15  lr: 0.000125  grad: 0.2108 (0.2342)  loss: 0.8977 (0.9058)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [5]  [3200/6250]  eta: 0:13:48  lr: 0.000125  grad: 0.1657 (0.2334)  loss: 0.9040 (0.9057)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [5]  [3300/6250]  eta: 0:13:27  lr: 0.000125  grad: 0.1904 (0.2336)  loss: 0.9011 (0.9055)  time: 0.6378  data: 0.3610  max mem: 26157
Train: [5]  [3400/6250]  eta: 0:13:00  lr: 0.000125  grad: 0.1918 (0.2329)  loss: 0.9045 (0.9054)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [5]  [3500/6250]  eta: 0:12:32  lr: 0.000125  grad: 0.1806 (0.2320)  loss: 0.9036 (0.9052)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [5]  [3600/6250]  eta: 0:12:05  lr: 0.000125  grad: 0.2107 (0.2322)  loss: 0.9019 (0.9050)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [5]  [3700/6250]  eta: 0:11:37  lr: 0.000125  grad: 0.1659 (0.2313)  loss: 0.8990 (0.9048)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [5]  [3800/6250]  eta: 0:11:10  lr: 0.000125  grad: 0.1657 (0.2306)  loss: 0.8990 (0.9047)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [5]  [3900/6250]  eta: 0:10:42  lr: 0.000125  grad: 0.1923 (0.2302)  loss: 0.8996 (0.9045)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [5]  [4000/6250]  eta: 0:10:14  lr: 0.000125  grad: 0.2156 (0.2302)  loss: 0.8985 (0.9044)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [5]  [4100/6250]  eta: 0:09:47  lr: 0.000125  grad: 0.1636 (0.2295)  loss: 0.8930 (0.9042)  time: 0.2777  data: 0.0064  max mem: 26157
Train: [5]  [4200/6250]  eta: 0:09:19  lr: 0.000125  grad: 0.1658 (0.2289)  loss: 0.8977 (0.9041)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [5]  [4300/6250]  eta: 0:08:52  lr: 0.000125  grad: 0.1582 (0.2281)  loss: 0.9014 (0.9039)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [5]  [4400/6250]  eta: 0:08:25  lr: 0.000125  grad: 0.1908 (0.2277)  loss: 0.8956 (0.9038)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [5]  [4500/6250]  eta: 0:07:57  lr: 0.000125  grad: 0.2406 (0.2273)  loss: 0.8964 (0.9036)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [5]  [4600/6250]  eta: 0:07:30  lr: 0.000125  grad: 0.2129 (0.2271)  loss: 0.8926 (0.9034)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [5]  [4700/6250]  eta: 0:07:02  lr: 0.000125  grad: 0.1836 (0.2266)  loss: 0.8928 (0.9032)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [5]  [4800/6250]  eta: 0:06:35  lr: 0.000125  grad: 0.1990 (0.2263)  loss: 0.8923 (0.9030)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [5]  [4900/6250]  eta: 0:06:08  lr: 0.000125  grad: 0.1561 (0.2259)  loss: 0.8916 (0.9028)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [5]  [5000/6250]  eta: 0:05:40  lr: 0.000125  grad: 0.2061 (0.2254)  loss: 0.8924 (0.9026)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [5]  [5100/6250]  eta: 0:05:13  lr: 0.000125  grad: 0.1846 (0.2248)  loss: 0.8920 (0.9024)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [5]  [5200/6250]  eta: 0:04:46  lr: 0.000125  grad: 0.2090 (0.2246)  loss: 0.8901 (0.9021)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [5]  [5300/6250]  eta: 0:04:18  lr: 0.000125  grad: 0.1999 (0.2242)  loss: 0.8887 (0.9019)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [5]  [5400/6250]  eta: 0:03:51  lr: 0.000125  grad: 0.1801 (0.2237)  loss: 0.8831 (0.9016)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [5]  [5500/6250]  eta: 0:03:24  lr: 0.000125  grad: 0.2115 (0.2235)  loss: 0.8904 (0.9014)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [5]  [5600/6250]  eta: 0:02:56  lr: 0.000125  grad: 0.2020 (0.2232)  loss: 0.8816 (0.9011)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [5]  [5700/6250]  eta: 0:02:29  lr: 0.000125  grad: 0.1903 (0.2229)  loss: 0.8836 (0.9008)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [5]  [5800/6250]  eta: 0:02:02  lr: 0.000125  grad: 0.1822 (0.2223)  loss: 0.8864 (0.9006)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [5]  [5900/6250]  eta: 0:01:35  lr: 0.000125  grad: 0.2036 (0.2222)  loss: 0.8838 (0.9003)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [5]  [6000/6250]  eta: 0:01:08  lr: 0.000125  grad: 0.2003 (0.2221)  loss: 0.8840 (0.9000)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [5]  [6100/6250]  eta: 0:00:40  lr: 0.000125  grad: 0.1967 (0.2217)  loss: 0.8809 (0.8998)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [5]  [6200/6250]  eta: 0:00:13  lr: 0.000125  grad: 0.1879 (0.2217)  loss: 0.8842 (0.8995)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [5]  [6249/6250]  eta: 0:00:00  lr: 0.000125  grad: 0.1824 (0.2214)  loss: 0.8819 (0.8994)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [5] Total time: 0:28:24 (0.2727 s / it)
Averaged stats: lr: 0.000125  grad: 0.1824 (0.2214)  loss: 0.8819 (0.8994)
Eval (hcp-train-subset): [5]  [ 0/62]  eta: 0:04:40  loss: 0.8934 (0.8934)  time: 4.5191  data: 4.4357  max mem: 26157
Eval (hcp-train-subset): [5]  [61/62]  eta: 0:00:00  loss: 0.8870 (0.8862)  time: 0.1307  data: 0.0477  max mem: 26157
Eval (hcp-train-subset): [5] Total time: 0:00:13 (0.2105 s / it)
Averaged stats (hcp-train-subset): loss: 0.8870 (0.8862)
Making plots (hcp-train-subset): example=38
Eval (hcp-val): [5]  [ 0/62]  eta: 0:05:45  loss: 0.8784 (0.8784)  time: 5.5769  data: 5.4928  max mem: 26157
Eval (hcp-val): [5]  [61/62]  eta: 0:00:00  loss: 0.8795 (0.8809)  time: 0.1193  data: 0.0364  max mem: 26157
Eval (hcp-val): [5] Total time: 0:00:13 (0.2104 s / it)
Averaged stats (hcp-val): loss: 0.8795 (0.8809)
Making plots (hcp-val): example=4
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [6]  [   0/6250]  eta: 9:00:55  lr: 0.000125  grad: 0.2315 (0.2315)  loss: 0.8599 (0.8599)  time: 5.1929  data: 4.8768  max mem: 26157
Train: [6]  [ 100/6250]  eta: 0:33:47  lr: 0.000125  grad: 0.1563 (0.1891)  loss: 0.8863 (0.8847)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [6]  [ 200/6250]  eta: 0:30:18  lr: 0.000125  grad: 0.1656 (0.2004)  loss: 0.8863 (0.8839)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [6]  [ 300/6250]  eta: 0:28:48  lr: 0.000125  grad: 0.1889 (0.1953)  loss: 0.8819 (0.8836)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [6]  [ 400/6250]  eta: 0:27:50  lr: 0.000125  grad: 0.1798 (0.1955)  loss: 0.8865 (0.8838)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [6]  [ 500/6250]  eta: 0:27:05  lr: 0.000125  grad: 0.1715 (0.1973)  loss: 0.8754 (0.8834)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [6]  [ 600/6250]  eta: 0:26:25  lr: 0.000125  grad: 0.1985 (0.1990)  loss: 0.8816 (0.8831)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [6]  [ 700/6250]  eta: 0:25:50  lr: 0.000125  grad: 0.1542 (0.1997)  loss: 0.8746 (0.8826)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [6]  [ 800/6250]  eta: 0:25:23  lr: 0.000125  grad: 0.1975 (0.2007)  loss: 0.8793 (0.8821)  time: 0.3120  data: 0.0379  max mem: 26157
Train: [6]  [ 900/6250]  eta: 0:24:51  lr: 0.000125  grad: 0.1742 (0.1990)  loss: 0.8771 (0.8819)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [6]  [1000/6250]  eta: 0:24:19  lr: 0.000125  grad: 0.1833 (0.1979)  loss: 0.8808 (0.8816)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [6]  [1100/6250]  eta: 0:23:49  lr: 0.000125  grad: 0.3059 (0.1987)  loss: 0.8774 (0.8813)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [6]  [1200/6250]  eta: 0:23:18  lr: 0.000125  grad: 0.1682 (0.1983)  loss: 0.8806 (0.8811)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [6]  [1300/6250]  eta: 0:22:53  lr: 0.000125  grad: 0.1492 (0.1969)  loss: 0.8787 (0.8811)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [6]  [1400/6250]  eta: 0:22:23  lr: 0.000125  grad: 0.2112 (0.1968)  loss: 0.8757 (0.8807)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [6]  [1500/6250]  eta: 0:21:53  lr: 0.000125  grad: 0.1788 (0.1957)  loss: 0.8755 (0.8805)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [6]  [1600/6250]  eta: 0:21:54  lr: 0.000125  grad: 0.1465 (0.1952)  loss: 0.8719 (0.8801)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [6]  [1700/6250]  eta: 0:21:23  lr: 0.000125  grad: 0.1740 (0.1940)  loss: 0.8782 (0.8800)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [6]  [1800/6250]  eta: 0:20:53  lr: 0.000125  grad: 0.1475 (0.1923)  loss: 0.8742 (0.8798)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [6]  [1900/6250]  eta: 0:20:22  lr: 0.000125  grad: 0.1689 (0.1916)  loss: 0.8750 (0.8795)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [6]  [2000/6250]  eta: 0:19:51  lr: 0.000125  grad: 0.1427 (0.1910)  loss: 0.8773 (0.8794)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [6]  [2100/6250]  eta: 0:19:21  lr: 0.000125  grad: 0.1568 (0.1898)  loss: 0.8760 (0.8792)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [6]  [2200/6250]  eta: 0:18:51  lr: 0.000125  grad: 0.1544 (0.1885)  loss: 0.8718 (0.8790)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [6]  [2300/6250]  eta: 0:18:22  lr: 0.000125  grad: 0.1623 (0.1882)  loss: 0.8724 (0.8787)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [6]  [2400/6250]  eta: 0:17:53  lr: 0.000125  grad: 0.1609 (0.1874)  loss: 0.8709 (0.8785)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [6]  [2500/6250]  eta: 0:17:24  lr: 0.000125  grad: 0.1482 (0.1864)  loss: 0.8682 (0.8783)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [6]  [2600/6250]  eta: 0:16:55  lr: 0.000125  grad: 0.1984 (0.1857)  loss: 0.8749 (0.8781)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [6]  [2700/6250]  eta: 0:16:26  lr: 0.000125  grad: 0.1729 (0.1851)  loss: 0.8793 (0.8780)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [6]  [2800/6250]  eta: 0:15:57  lr: 0.000125  grad: 0.1597 (0.1846)  loss: 0.8690 (0.8778)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [6]  [2900/6250]  eta: 0:15:28  lr: 0.000125  grad: 0.1572 (0.1843)  loss: 0.8725 (0.8776)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [6]  [3000/6250]  eta: 0:15:00  lr: 0.000125  grad: 0.1391 (0.1837)  loss: 0.8729 (0.8775)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [6]  [3100/6250]  eta: 0:14:31  lr: 0.000125  grad: 0.1414 (0.1829)  loss: 0.8746 (0.8773)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [6]  [3200/6250]  eta: 0:14:03  lr: 0.000125  grad: 0.1660 (0.1821)  loss: 0.8722 (0.8771)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [6]  [3300/6250]  eta: 0:13:35  lr: 0.000125  grad: 0.1427 (0.1814)  loss: 0.8710 (0.8769)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [6]  [3400/6250]  eta: 0:13:06  lr: 0.000125  grad: 0.1518 (0.1806)  loss: 0.8715 (0.8767)  time: 0.2766  data: 0.0002  max mem: 26157
Train: [6]  [3500/6250]  eta: 0:12:38  lr: 0.000125  grad: 0.1417 (0.1799)  loss: 0.8676 (0.8765)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [6]  [3600/6250]  eta: 0:12:10  lr: 0.000125  grad: 0.1195 (0.1792)  loss: 0.8706 (0.8763)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [6]  [3700/6250]  eta: 0:11:42  lr: 0.000125  grad: 0.1406 (0.1782)  loss: 0.8635 (0.8761)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [6]  [3800/6250]  eta: 0:11:15  lr: 0.000125  grad: 0.1700 (0.1776)  loss: 0.8648 (0.8759)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [6]  [3900/6250]  eta: 0:10:47  lr: 0.000125  grad: 0.1372 (0.1767)  loss: 0.8684 (0.8758)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [6]  [4000/6250]  eta: 0:10:19  lr: 0.000125  grad: 0.1435 (0.1761)  loss: 0.8699 (0.8756)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [6]  [4100/6250]  eta: 0:09:51  lr: 0.000125  grad: 0.1460 (0.1754)  loss: 0.8696 (0.8755)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [6]  [4200/6250]  eta: 0:09:23  lr: 0.000125  grad: 0.1522 (0.1747)  loss: 0.8713 (0.8753)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [6]  [4300/6250]  eta: 0:08:56  lr: 0.000125  grad: 0.1248 (0.1738)  loss: 0.8676 (0.8751)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [6]  [4400/6250]  eta: 0:08:28  lr: 0.000125  grad: 0.1215 (0.1730)  loss: 0.8715 (0.8750)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [6]  [4500/6250]  eta: 0:08:00  lr: 0.000125  grad: 0.1374 (0.1723)  loss: 0.8657 (0.8748)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [6]  [4600/6250]  eta: 0:07:33  lr: 0.000125  grad: 0.1199 (0.1717)  loss: 0.8633 (0.8746)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [6]  [4700/6250]  eta: 0:07:05  lr: 0.000125  grad: 0.1373 (0.1712)  loss: 0.8666 (0.8745)  time: 0.2699  data: 0.0001  max mem: 26157
Train: [6]  [4800/6250]  eta: 0:06:40  lr: 0.000125  grad: 0.1600 (0.1708)  loss: 0.8684 (0.8743)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [6]  [4900/6250]  eta: 0:06:12  lr: 0.000125  grad: 0.1405 (0.1702)  loss: 0.8660 (0.8741)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [6]  [5000/6250]  eta: 0:05:44  lr: 0.000125  grad: 0.1232 (0.1700)  loss: 0.8624 (0.8739)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [6]  [5100/6250]  eta: 0:05:17  lr: 0.000125  grad: 0.1404 (0.1694)  loss: 0.8609 (0.8737)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [6]  [5200/6250]  eta: 0:04:49  lr: 0.000125  grad: 0.1359 (0.1692)  loss: 0.8625 (0.8734)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [6]  [5300/6250]  eta: 0:04:22  lr: 0.000125  grad: 0.1545 (0.1685)  loss: 0.8631 (0.8732)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [6]  [5400/6250]  eta: 0:03:54  lr: 0.000125  grad: 0.1291 (0.1679)  loss: 0.8611 (0.8730)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [6]  [5500/6250]  eta: 0:03:27  lr: 0.000125  grad: 0.1428 (0.1673)  loss: 0.8591 (0.8728)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [6]  [5600/6250]  eta: 0:02:59  lr: 0.000125  grad: 0.1391 (0.1667)  loss: 0.8606 (0.8726)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [6]  [5700/6250]  eta: 0:02:31  lr: 0.000125  grad: 0.1114 (0.1660)  loss: 0.8625 (0.8725)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [6]  [5800/6250]  eta: 0:02:04  lr: 0.000125  grad: 0.1206 (0.1652)  loss: 0.8641 (0.8723)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [6]  [5900/6250]  eta: 0:01:36  lr: 0.000125  grad: 0.1250 (0.1647)  loss: 0.8614 (0.8721)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [6]  [6000/6250]  eta: 0:01:08  lr: 0.000125  grad: 0.1294 (0.1640)  loss: 0.8616 (0.8720)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [6]  [6100/6250]  eta: 0:00:41  lr: 0.000125  grad: 0.1430 (0.1638)  loss: 0.8601 (0.8718)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [6]  [6200/6250]  eta: 0:00:13  lr: 0.000125  grad: 0.1116 (0.1631)  loss: 0.8605 (0.8717)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [6]  [6249/6250]  eta: 0:00:00  lr: 0.000125  grad: 0.1026 (0.1627)  loss: 0.8609 (0.8716)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [6] Total time: 0:28:48 (0.2766 s / it)
Averaged stats: lr: 0.000125  grad: 0.1026 (0.1627)  loss: 0.8609 (0.8716)
Eval (hcp-train-subset): [6]  [ 0/62]  eta: 0:04:25  loss: 0.8743 (0.8743)  time: 4.2889  data: 4.2045  max mem: 26157
Eval (hcp-train-subset): [6]  [61/62]  eta: 0:00:00  loss: 0.8675 (0.8668)  time: 0.1291  data: 0.0459  max mem: 26157
Eval (hcp-train-subset): [6] Total time: 0:00:12 (0.2081 s / it)
Averaged stats (hcp-train-subset): loss: 0.8675 (0.8668)
Making plots (hcp-train-subset): example=18
Eval (hcp-val): [6]  [ 0/62]  eta: 0:05:15  loss: 0.8591 (0.8591)  time: 5.0889  data: 5.0047  max mem: 26157
Eval (hcp-val): [6]  [61/62]  eta: 0:00:00  loss: 0.8605 (0.8614)  time: 0.1295  data: 0.0450  max mem: 26157
Eval (hcp-val): [6] Total time: 0:00:12 (0.2090 s / it)
Averaged stats (hcp-val): loss: 0.8605 (0.8614)
Making plots (hcp-val): example=58
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [7]  [   0/6250]  eta: 10:44:37  lr: 0.000125  grad: 0.3638 (0.3638)  loss: 0.8120 (0.8120)  time: 6.1883  data: 5.8908  max mem: 26157
Train: [7]  [ 100/6250]  eta: 0:33:53  lr: 0.000125  grad: 0.1102 (0.1370)  loss: 0.8551 (0.8560)  time: 0.2747  data: 0.0002  max mem: 26157
Train: [7]  [ 200/6250]  eta: 0:30:18  lr: 0.000125  grad: 0.1280 (0.1433)  loss: 0.8558 (0.8550)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [7]  [ 300/6250]  eta: 0:28:48  lr: 0.000125  grad: 0.1206 (0.1455)  loss: 0.8558 (0.8547)  time: 0.2709  data: 0.0003  max mem: 26157
Train: [7]  [ 400/6250]  eta: 0:27:49  lr: 0.000125  grad: 0.1324 (0.1441)  loss: 0.8557 (0.8548)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [7]  [ 500/6250]  eta: 0:27:02  lr: 0.000125  grad: 0.1433 (0.1429)  loss: 0.8623 (0.8552)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [7]  [ 600/6250]  eta: 0:26:21  lr: 0.000125  grad: 0.1215 (0.1398)  loss: 0.8597 (0.8557)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [7]  [ 700/6250]  eta: 0:25:44  lr: 0.000125  grad: 0.1493 (0.1404)  loss: 0.8583 (0.8561)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [7]  [ 800/6250]  eta: 0:25:10  lr: 0.000125  grad: 0.1054 (0.1389)  loss: 0.8586 (0.8563)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [7]  [ 900/6250]  eta: 0:24:37  lr: 0.000125  grad: 0.1162 (0.1379)  loss: 0.8565 (0.8564)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [7]  [1000/6250]  eta: 0:24:06  lr: 0.000125  grad: 0.1190 (0.1370)  loss: 0.8562 (0.8565)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [7]  [1100/6250]  eta: 0:23:35  lr: 0.000125  grad: 0.1218 (0.1356)  loss: 0.8572 (0.8565)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [7]  [1200/6250]  eta: 0:23:05  lr: 0.000125  grad: 0.1023 (0.1344)  loss: 0.8611 (0.8566)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [7]  [1300/6250]  eta: 0:22:35  lr: 0.000125  grad: 0.1130 (0.1340)  loss: 0.8587 (0.8566)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [7]  [1400/6250]  eta: 0:22:06  lr: 0.000125  grad: 0.1160 (0.1338)  loss: 0.8560 (0.8567)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [7]  [1500/6250]  eta: 0:21:38  lr: 0.000125  grad: 0.1232 (0.1332)  loss: 0.8586 (0.8567)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [7]  [1600/6250]  eta: 0:21:09  lr: 0.000125  grad: 0.1150 (0.1330)  loss: 0.8547 (0.8567)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [7]  [1700/6250]  eta: 0:20:41  lr: 0.000125  grad: 0.1112 (0.1322)  loss: 0.8547 (0.8566)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [7]  [1800/6250]  eta: 0:20:13  lr: 0.000125  grad: 0.1208 (0.1315)  loss: 0.8552 (0.8566)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [7]  [1900/6250]  eta: 0:19:45  lr: 0.000125  grad: 0.1094 (0.1312)  loss: 0.8559 (0.8567)  time: 0.2705  data: 0.0011  max mem: 26157
Train: [7]  [2000/6250]  eta: 0:19:18  lr: 0.000125  grad: 0.1353 (0.1311)  loss: 0.8546 (0.8566)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [7]  [2100/6250]  eta: 0:18:50  lr: 0.000125  grad: 0.1104 (0.1308)  loss: 0.8520 (0.8565)  time: 0.2776  data: 0.0002  max mem: 26157
Train: [7]  [2200/6250]  eta: 0:18:23  lr: 0.000125  grad: 0.1176 (0.1306)  loss: 0.8588 (0.8564)  time: 0.2845  data: 0.0003  max mem: 26157
Train: [7]  [2300/6250]  eta: 0:18:02  lr: 0.000125  grad: 0.1348 (0.1305)  loss: 0.8532 (0.8564)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [7]  [2400/6250]  eta: 0:17:34  lr: 0.000125  grad: 0.1156 (0.1304)  loss: 0.8543 (0.8563)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [7]  [2500/6250]  eta: 0:17:06  lr: 0.000125  grad: 0.1248 (0.1299)  loss: 0.8550 (0.8562)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [7]  [2600/6250]  eta: 0:16:38  lr: 0.000125  grad: 0.1130 (0.1293)  loss: 0.8581 (0.8563)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [7]  [2700/6250]  eta: 0:16:10  lr: 0.000125  grad: 0.1117 (0.1288)  loss: 0.8534 (0.8562)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [7]  [2800/6250]  eta: 0:15:42  lr: 0.000125  grad: 0.1058 (0.1285)  loss: 0.8550 (0.8562)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [7]  [2900/6250]  eta: 0:15:14  lr: 0.000125  grad: 0.1163 (0.1282)  loss: 0.8584 (0.8562)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [7]  [3000/6250]  eta: 0:14:46  lr: 0.000125  grad: 0.1041 (0.1278)  loss: 0.8569 (0.8562)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [7]  [3100/6250]  eta: 0:14:19  lr: 0.000125  grad: 0.0986 (0.1272)  loss: 0.8558 (0.8562)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [7]  [3200/6250]  eta: 0:13:51  lr: 0.000125  grad: 0.1154 (0.1268)  loss: 0.8559 (0.8562)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [7]  [3300/6250]  eta: 0:13:23  lr: 0.000125  grad: 0.0989 (0.1264)  loss: 0.8579 (0.8562)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [7]  [3400/6250]  eta: 0:12:56  lr: 0.000125  grad: 0.0955 (0.1259)  loss: 0.8525 (0.8562)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [7]  [3500/6250]  eta: 0:12:29  lr: 0.000125  grad: 0.0935 (0.1256)  loss: 0.8559 (0.8562)  time: 0.3128  data: 0.0002  max mem: 26157
Train: [7]  [3600/6250]  eta: 0:12:01  lr: 0.000125  grad: 0.1058 (0.1250)  loss: 0.8543 (0.8562)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [7]  [3700/6250]  eta: 0:11:34  lr: 0.000125  grad: 0.1133 (0.1246)  loss: 0.8590 (0.8562)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [7]  [3800/6250]  eta: 0:11:06  lr: 0.000125  grad: 0.1027 (0.1242)  loss: 0.8566 (0.8562)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [7]  [3900/6250]  eta: 0:10:39  lr: 0.000125  grad: 0.1058 (0.1239)  loss: 0.8563 (0.8562)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [7]  [4000/6250]  eta: 0:10:12  lr: 0.000125  grad: 0.0948 (0.1234)  loss: 0.8572 (0.8562)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [7]  [4100/6250]  eta: 0:09:44  lr: 0.000125  grad: 0.0957 (0.1229)  loss: 0.8594 (0.8562)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [7]  [4200/6250]  eta: 0:09:17  lr: 0.000125  grad: 0.1097 (0.1226)  loss: 0.8556 (0.8561)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [7]  [4300/6250]  eta: 0:08:50  lr: 0.000125  grad: 0.0947 (0.1222)  loss: 0.8534 (0.8561)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [7]  [4400/6250]  eta: 0:08:22  lr: 0.000125  grad: 0.0963 (0.1218)  loss: 0.8576 (0.8561)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [7]  [4500/6250]  eta: 0:07:55  lr: 0.000125  grad: 0.1126 (0.1216)  loss: 0.8562 (0.8561)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [7]  [4600/6250]  eta: 0:07:28  lr: 0.000125  grad: 0.0931 (0.1214)  loss: 0.8571 (0.8561)  time: 0.2769  data: 0.0003  max mem: 26157
Train: [7]  [4700/6250]  eta: 0:07:01  lr: 0.000125  grad: 0.1003 (0.1210)  loss: 0.8565 (0.8561)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [7]  [4800/6250]  eta: 0:06:34  lr: 0.000125  grad: 0.1087 (0.1207)  loss: 0.8531 (0.8562)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [7]  [4900/6250]  eta: 0:06:06  lr: 0.000125  grad: 0.1041 (0.1204)  loss: 0.8578 (0.8561)  time: 0.2749  data: 0.0002  max mem: 26157
Train: [7]  [5000/6250]  eta: 0:05:39  lr: 0.000125  grad: 0.0958 (0.1200)  loss: 0.8506 (0.8561)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [7]  [5100/6250]  eta: 0:05:12  lr: 0.000125  grad: 0.0881 (0.1196)  loss: 0.8564 (0.8561)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [7]  [5200/6250]  eta: 0:04:45  lr: 0.000125  grad: 0.0995 (0.1193)  loss: 0.8541 (0.8560)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [7]  [5300/6250]  eta: 0:04:18  lr: 0.000125  grad: 0.0876 (0.1189)  loss: 0.8542 (0.8560)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [7]  [5400/6250]  eta: 0:03:50  lr: 0.000125  grad: 0.0973 (0.1186)  loss: 0.8571 (0.8560)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [7]  [5500/6250]  eta: 0:03:23  lr: 0.000125  grad: 0.0934 (0.1183)  loss: 0.8558 (0.8560)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [7]  [5600/6250]  eta: 0:02:56  lr: 0.000125  grad: 0.1005 (0.1181)  loss: 0.8524 (0.8560)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [7]  [5700/6250]  eta: 0:02:29  lr: 0.000125  grad: 0.0869 (0.1177)  loss: 0.8567 (0.8559)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [7]  [5800/6250]  eta: 0:02:02  lr: 0.000125  grad: 0.0968 (0.1176)  loss: 0.8556 (0.8559)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [7]  [5900/6250]  eta: 0:01:35  lr: 0.000125  grad: 0.0889 (0.1174)  loss: 0.8538 (0.8559)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [7]  [6000/6250]  eta: 0:01:07  lr: 0.000125  grad: 0.0953 (0.1173)  loss: 0.8537 (0.8558)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [7]  [6100/6250]  eta: 0:00:40  lr: 0.000125  grad: 0.0933 (0.1170)  loss: 0.8540 (0.8558)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [7]  [6200/6250]  eta: 0:00:13  lr: 0.000125  grad: 0.0834 (0.1166)  loss: 0.8547 (0.8558)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [7]  [6249/6250]  eta: 0:00:00  lr: 0.000125  grad: 0.1022 (0.1165)  loss: 0.8546 (0.8558)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [7] Total time: 0:28:23 (0.2725 s / it)
Averaged stats: lr: 0.000125  grad: 0.1022 (0.1165)  loss: 0.8546 (0.8558)
Eval (hcp-train-subset): [7]  [ 0/62]  eta: 0:02:44  loss: 0.8668 (0.8668)  time: 2.6568  data: 2.5544  max mem: 26157
Eval (hcp-train-subset): [7]  [61/62]  eta: 0:00:00  loss: 0.8580 (0.8589)  time: 0.1385  data: 0.0559  max mem: 26157
Eval (hcp-train-subset): [7] Total time: 0:00:11 (0.1931 s / it)
Averaged stats (hcp-train-subset): loss: 0.8580 (0.8589)
Making plots (hcp-train-subset): example=25
Eval (hcp-val): [7]  [ 0/62]  eta: 0:03:37  loss: 0.8464 (0.8464)  time: 3.5049  data: 3.3752  max mem: 26157
Eval (hcp-val): [7]  [61/62]  eta: 0:00:00  loss: 0.8523 (0.8533)  time: 0.1248  data: 0.0405  max mem: 26157
Eval (hcp-val): [7] Total time: 0:00:11 (0.1911 s / it)
Averaged stats (hcp-val): loss: 0.8523 (0.8533)
Making plots (hcp-val): example=50
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [8]  [   0/6250]  eta: 9:47:03  lr: 0.000125  grad: 0.1016 (0.1016)  loss: 0.8817 (0.8817)  time: 5.6358  data: 5.3595  max mem: 26157
Train: [8]  [ 100/6250]  eta: 0:32:57  lr: 0.000125  grad: 0.1063 (0.1091)  loss: 0.8495 (0.8569)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [8]  [ 200/6250]  eta: 0:29:50  lr: 0.000125  grad: 0.0918 (0.1074)  loss: 0.8511 (0.8543)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [8]  [ 300/6250]  eta: 0:28:30  lr: 0.000125  grad: 0.0973 (0.1056)  loss: 0.8503 (0.8530)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [8]  [ 400/6250]  eta: 0:27:35  lr: 0.000125  grad: 0.0969 (0.1044)  loss: 0.8468 (0.8525)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [8]  [ 500/6250]  eta: 0:26:51  lr: 0.000125  grad: 0.0994 (0.1052)  loss: 0.8421 (0.8516)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [8]  [ 600/6250]  eta: 0:27:04  lr: 0.000125  grad: 0.0863 (0.1057)  loss: 0.8443 (0.8512)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [8]  [ 700/6250]  eta: 0:26:21  lr: 0.000125  grad: 0.0930 (0.1041)  loss: 0.8526 (0.8509)  time: 0.2702  data: 0.0003  max mem: 26157
Train: [8]  [ 800/6250]  eta: 0:25:42  lr: 0.000125  grad: 0.0937 (0.1030)  loss: 0.8497 (0.8509)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [8]  [ 900/6250]  eta: 0:25:05  lr: 0.000125  grad: 0.0968 (0.1019)  loss: 0.8533 (0.8511)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [8]  [1000/6250]  eta: 0:24:30  lr: 0.000125  grad: 0.0889 (0.1020)  loss: 0.8548 (0.8514)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [8]  [1100/6250]  eta: 0:23:57  lr: 0.000125  grad: 0.0904 (0.1013)  loss: 0.8524 (0.8516)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [8]  [1200/6250]  eta: 0:23:24  lr: 0.000125  grad: 0.0887 (0.1011)  loss: 0.8483 (0.8516)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [8]  [1300/6250]  eta: 0:22:53  lr: 0.000125  grad: 0.0855 (0.1006)  loss: 0.8529 (0.8517)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [8]  [1400/6250]  eta: 0:22:23  lr: 0.000125  grad: 0.0909 (0.1006)  loss: 0.8511 (0.8519)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [8]  [1500/6250]  eta: 0:21:53  lr: 0.000125  grad: 0.0884 (0.0999)  loss: 0.8522 (0.8519)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [8]  [1600/6250]  eta: 0:21:23  lr: 0.000125  grad: 0.1031 (0.0997)  loss: 0.8530 (0.8520)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [8]  [1700/6250]  eta: 0:20:54  lr: 0.000125  grad: 0.0983 (0.0994)  loss: 0.8526 (0.8521)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [8]  [1800/6250]  eta: 0:20:26  lr: 0.000125  grad: 0.0968 (0.0997)  loss: 0.8494 (0.8522)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [8]  [1900/6250]  eta: 0:19:57  lr: 0.000125  grad: 0.0943 (0.0994)  loss: 0.8548 (0.8523)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [8]  [2000/6250]  eta: 0:19:31  lr: 0.000125  grad: 0.0885 (0.0991)  loss: 0.8538 (0.8523)  time: 0.3233  data: 0.0520  max mem: 26157
Train: [8]  [2100/6250]  eta: 0:19:03  lr: 0.000125  grad: 0.0811 (0.0989)  loss: 0.8495 (0.8521)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [8]  [2200/6250]  eta: 0:18:34  lr: 0.000125  grad: 0.0939 (0.0987)  loss: 0.8522 (0.8521)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [8]  [2300/6250]  eta: 0:18:06  lr: 0.000125  grad: 0.0955 (0.0986)  loss: 0.8487 (0.8520)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [8]  [2400/6250]  eta: 0:17:37  lr: 0.000125  grad: 0.0910 (0.0985)  loss: 0.8509 (0.8521)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [8]  [2500/6250]  eta: 0:17:09  lr: 0.000125  grad: 0.0917 (0.0982)  loss: 0.8531 (0.8521)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [8]  [2600/6250]  eta: 0:16:41  lr: 0.000125  grad: 0.0847 (0.0979)  loss: 0.8496 (0.8522)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [8]  [2700/6250]  eta: 0:16:13  lr: 0.000125  grad: 0.0891 (0.0979)  loss: 0.8533 (0.8522)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [8]  [2800/6250]  eta: 0:15:45  lr: 0.000125  grad: 0.0942 (0.0978)  loss: 0.8516 (0.8522)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [8]  [2900/6250]  eta: 0:15:17  lr: 0.000125  grad: 0.0865 (0.0976)  loss: 0.8564 (0.8522)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [8]  [3000/6250]  eta: 0:14:49  lr: 0.000125  grad: 0.0785 (0.0973)  loss: 0.8560 (0.8522)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [8]  [3100/6250]  eta: 0:14:21  lr: 0.000125  grad: 0.0966 (0.0972)  loss: 0.8536 (0.8522)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [8]  [3200/6250]  eta: 0:13:54  lr: 0.000125  grad: 0.0877 (0.0970)  loss: 0.8562 (0.8521)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [8]  [3300/6250]  eta: 0:13:26  lr: 0.000125  grad: 0.0907 (0.0968)  loss: 0.8539 (0.8521)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [8]  [3400/6250]  eta: 0:12:58  lr: 0.000125  grad: 0.0876 (0.0968)  loss: 0.8506 (0.8521)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [8]  [3500/6250]  eta: 0:12:31  lr: 0.000125  grad: 0.0831 (0.0965)  loss: 0.8569 (0.8521)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [8]  [3600/6250]  eta: 0:12:05  lr: 0.000125  grad: 0.0848 (0.0963)  loss: 0.8503 (0.8521)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [8]  [3700/6250]  eta: 0:11:37  lr: 0.000125  grad: 0.0860 (0.0961)  loss: 0.8522 (0.8521)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [8]  [3800/6250]  eta: 0:11:09  lr: 0.000125  grad: 0.0812 (0.0958)  loss: 0.8552 (0.8522)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [8]  [3900/6250]  eta: 0:10:42  lr: 0.000125  grad: 0.0837 (0.0957)  loss: 0.8538 (0.8521)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [8]  [4000/6250]  eta: 0:10:14  lr: 0.000125  grad: 0.0909 (0.0955)  loss: 0.8517 (0.8521)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [8]  [4100/6250]  eta: 0:09:47  lr: 0.000125  grad: 0.0815 (0.0953)  loss: 0.8528 (0.8521)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [8]  [4200/6250]  eta: 0:09:19  lr: 0.000125  grad: 0.0815 (0.0951)  loss: 0.8515 (0.8521)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [8]  [4300/6250]  eta: 0:08:52  lr: 0.000125  grad: 0.0796 (0.0949)  loss: 0.8568 (0.8521)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [8]  [4400/6250]  eta: 0:08:24  lr: 0.000125  grad: 0.0853 (0.0947)  loss: 0.8516 (0.8521)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [8]  [4500/6250]  eta: 0:07:57  lr: 0.000125  grad: 0.0892 (0.0946)  loss: 0.8523 (0.8521)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [8]  [4600/6250]  eta: 0:07:30  lr: 0.000125  grad: 0.0772 (0.0944)  loss: 0.8502 (0.8521)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [8]  [4700/6250]  eta: 0:07:02  lr: 0.000125  grad: 0.0814 (0.0943)  loss: 0.8498 (0.8521)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [8]  [4800/6250]  eta: 0:06:35  lr: 0.000125  grad: 0.0889 (0.0942)  loss: 0.8481 (0.8521)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [8]  [4900/6250]  eta: 0:06:07  lr: 0.000125  grad: 0.0783 (0.0940)  loss: 0.8513 (0.8521)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [8]  [5000/6250]  eta: 0:05:40  lr: 0.000125  grad: 0.0850 (0.0939)  loss: 0.8500 (0.8521)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [8]  [5100/6250]  eta: 0:05:13  lr: 0.000125  grad: 0.0883 (0.0940)  loss: 0.8537 (0.8520)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [8]  [5200/6250]  eta: 0:04:45  lr: 0.000124  grad: 0.0835 (0.0938)  loss: 0.8455 (0.8520)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [8]  [5300/6250]  eta: 0:04:18  lr: 0.000124  grad: 0.0911 (0.0936)  loss: 0.8488 (0.8520)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [8]  [5400/6250]  eta: 0:03:51  lr: 0.000124  grad: 0.0891 (0.0936)  loss: 0.8511 (0.8520)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [8]  [5500/6250]  eta: 0:03:24  lr: 0.000124  grad: 0.0826 (0.0935)  loss: 0.8474 (0.8519)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [8]  [5600/6250]  eta: 0:02:56  lr: 0.000124  grad: 0.0823 (0.0934)  loss: 0.8501 (0.8519)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [8]  [5700/6250]  eta: 0:02:29  lr: 0.000124  grad: 0.0837 (0.0932)  loss: 0.8482 (0.8518)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [8]  [5800/6250]  eta: 0:02:02  lr: 0.000124  grad: 0.0812 (0.0931)  loss: 0.8476 (0.8518)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [8]  [5900/6250]  eta: 0:01:35  lr: 0.000124  grad: 0.0878 (0.0930)  loss: 0.8524 (0.8518)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [8]  [6000/6250]  eta: 0:01:08  lr: 0.000124  grad: 0.0788 (0.0928)  loss: 0.8505 (0.8518)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [8]  [6100/6250]  eta: 0:00:40  lr: 0.000124  grad: 0.0844 (0.0927)  loss: 0.8497 (0.8517)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [8]  [6200/6250]  eta: 0:00:13  lr: 0.000124  grad: 0.0781 (0.0926)  loss: 0.8506 (0.8517)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [8]  [6249/6250]  eta: 0:00:00  lr: 0.000124  grad: 0.0810 (0.0925)  loss: 0.8461 (0.8516)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [8] Total time: 0:28:26 (0.2730 s / it)
Averaged stats: lr: 0.000124  grad: 0.0810 (0.0925)  loss: 0.8461 (0.8516)
Eval (hcp-train-subset): [8]  [ 0/62]  eta: 0:03:58  loss: 0.8651 (0.8651)  time: 3.8543  data: 3.7250  max mem: 26157
Eval (hcp-train-subset): [8]  [61/62]  eta: 0:00:00  loss: 0.8499 (0.8528)  time: 0.1403  data: 0.0576  max mem: 26157
Eval (hcp-train-subset): [8] Total time: 0:00:13 (0.2171 s / it)
Averaged stats (hcp-train-subset): loss: 0.8499 (0.8528)
Making plots (hcp-train-subset): example=8
Eval (hcp-val): [8]  [ 0/62]  eta: 0:04:43  loss: 0.8422 (0.8422)  time: 4.5720  data: 4.4882  max mem: 26157
Eval (hcp-val): [8]  [61/62]  eta: 0:00:00  loss: 0.8459 (0.8472)  time: 0.1243  data: 0.0414  max mem: 26157
Eval (hcp-val): [8] Total time: 0:00:12 (0.2011 s / it)
Averaged stats (hcp-val): loss: 0.8459 (0.8472)
Making plots (hcp-val): example=36
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [9]  [   0/6250]  eta: 9:18:35  lr: 0.000124  grad: 0.0533 (0.0533)  loss: 0.8792 (0.8792)  time: 5.3624  data: 5.0854  max mem: 26157
Train: [9]  [ 100/6250]  eta: 0:33:54  lr: 0.000124  grad: 0.0910 (0.1011)  loss: 0.8466 (0.8490)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [9]  [ 200/6250]  eta: 0:30:26  lr: 0.000124  grad: 0.0852 (0.0963)  loss: 0.8493 (0.8492)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [9]  [ 300/6250]  eta: 0:28:59  lr: 0.000124  grad: 0.0769 (0.0945)  loss: 0.8445 (0.8489)  time: 0.2765  data: 0.0002  max mem: 26157
Train: [9]  [ 400/6250]  eta: 0:28:00  lr: 0.000124  grad: 0.0858 (0.0940)  loss: 0.8493 (0.8486)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [9]  [ 500/6250]  eta: 0:27:14  lr: 0.000124  grad: 0.0905 (0.0934)  loss: 0.8407 (0.8481)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [9]  [ 600/6250]  eta: 0:27:17  lr: 0.000124  grad: 0.0810 (0.0925)  loss: 0.8456 (0.8481)  time: 0.2760  data: 0.0002  max mem: 26157
Train: [9]  [ 700/6250]  eta: 0:27:12  lr: 0.000124  grad: 0.0900 (0.0920)  loss: 0.8449 (0.8478)  time: 0.2747  data: 0.0002  max mem: 26157
Train: [9]  [ 800/6250]  eta: 0:26:28  lr: 0.000124  grad: 0.0826 (0.0916)  loss: 0.8431 (0.8475)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [9]  [ 900/6250]  eta: 0:25:50  lr: 0.000124  grad: 0.0830 (0.0910)  loss: 0.8460 (0.8475)  time: 0.2719  data: 0.0003  max mem: 26157
Train: [9]  [1000/6250]  eta: 0:25:10  lr: 0.000124  grad: 0.0808 (0.0903)  loss: 0.8521 (0.8476)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [9]  [1100/6250]  eta: 0:24:33  lr: 0.000124  grad: 0.0830 (0.0898)  loss: 0.8469 (0.8476)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [9]  [1200/6250]  eta: 0:23:57  lr: 0.000124  grad: 0.0803 (0.0893)  loss: 0.8511 (0.8475)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [9]  [1300/6250]  eta: 0:23:46  lr: 0.000124  grad: 0.0838 (0.0893)  loss: 0.8491 (0.8477)  time: 0.5207  data: 0.2471  max mem: 26157
Train: [9]  [1400/6250]  eta: 0:23:11  lr: 0.000124  grad: 0.0767 (0.0886)  loss: 0.8513 (0.8478)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [9]  [1500/6250]  eta: 0:22:37  lr: 0.000124  grad: 0.0817 (0.0885)  loss: 0.8423 (0.8478)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [9]  [1600/6250]  eta: 0:22:03  lr: 0.000124  grad: 0.0779 (0.0880)  loss: 0.8464 (0.8478)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [9]  [1700/6250]  eta: 0:21:30  lr: 0.000124  grad: 0.0765 (0.0878)  loss: 0.8473 (0.8478)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [9]  [1800/6250]  eta: 0:20:59  lr: 0.000124  grad: 0.0792 (0.0876)  loss: 0.8474 (0.8479)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [9]  [1900/6250]  eta: 0:20:27  lr: 0.000124  grad: 0.0809 (0.0873)  loss: 0.8510 (0.8478)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [9]  [2000/6250]  eta: 0:19:56  lr: 0.000124  grad: 0.0840 (0.0872)  loss: 0.8522 (0.8479)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [9]  [2100/6250]  eta: 0:19:26  lr: 0.000124  grad: 0.0888 (0.0870)  loss: 0.8470 (0.8479)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [9]  [2200/6250]  eta: 0:18:56  lr: 0.000124  grad: 0.0799 (0.0870)  loss: 0.8465 (0.8478)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [9]  [2300/6250]  eta: 0:18:25  lr: 0.000124  grad: 0.0827 (0.0867)  loss: 0.8450 (0.8478)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [9]  [2400/6250]  eta: 0:17:56  lr: 0.000124  grad: 0.0861 (0.0866)  loss: 0.8443 (0.8477)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [9]  [2500/6250]  eta: 0:17:26  lr: 0.000124  grad: 0.0832 (0.0866)  loss: 0.8454 (0.8476)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [9]  [2600/6250]  eta: 0:16:57  lr: 0.000124  grad: 0.0828 (0.0865)  loss: 0.8527 (0.8476)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [9]  [2700/6250]  eta: 0:16:28  lr: 0.000124  grad: 0.0805 (0.0865)  loss: 0.8483 (0.8475)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [9]  [2800/6250]  eta: 0:15:59  lr: 0.000124  grad: 0.0816 (0.0865)  loss: 0.8439 (0.8473)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [9]  [2900/6250]  eta: 0:15:30  lr: 0.000124  grad: 0.0846 (0.0865)  loss: 0.8451 (0.8472)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [9]  [3000/6250]  eta: 0:15:01  lr: 0.000124  grad: 0.0926 (0.0865)  loss: 0.8399 (0.8470)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [9]  [3100/6250]  eta: 0:14:32  lr: 0.000124  grad: 0.0836 (0.0869)  loss: 0.8442 (0.8469)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [9]  [3200/6250]  eta: 0:14:04  lr: 0.000124  grad: 0.0817 (0.0869)  loss: 0.8433 (0.8468)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [9]  [3300/6250]  eta: 0:13:36  lr: 0.000124  grad: 0.0910 (0.0869)  loss: 0.8429 (0.8467)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [9]  [3400/6250]  eta: 0:13:08  lr: 0.000124  grad: 0.0795 (0.0868)  loss: 0.8399 (0.8465)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [9]  [3500/6250]  eta: 0:12:39  lr: 0.000124  grad: 0.0773 (0.0867)  loss: 0.8410 (0.8464)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [9]  [3600/6250]  eta: 0:12:11  lr: 0.000124  grad: 0.0810 (0.0866)  loss: 0.8464 (0.8464)  time: 0.2719  data: 0.0003  max mem: 26157
Train: [9]  [3700/6250]  eta: 0:11:43  lr: 0.000124  grad: 0.0821 (0.0866)  loss: 0.8441 (0.8463)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [9]  [3800/6250]  eta: 0:11:15  lr: 0.000124  grad: 0.0851 (0.0864)  loss: 0.8445 (0.8463)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [9]  [3900/6250]  eta: 0:10:48  lr: 0.000124  grad: 0.0820 (0.0863)  loss: 0.8470 (0.8463)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [9]  [4000/6250]  eta: 0:10:20  lr: 0.000124  grad: 0.0873 (0.0862)  loss: 0.8427 (0.8462)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [9]  [4100/6250]  eta: 0:09:52  lr: 0.000124  grad: 0.0732 (0.0861)  loss: 0.8472 (0.8463)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [9]  [4200/6250]  eta: 0:09:24  lr: 0.000124  grad: 0.0738 (0.0860)  loss: 0.8489 (0.8463)  time: 0.2724  data: 0.0003  max mem: 26157
Train: [9]  [4300/6250]  eta: 0:08:56  lr: 0.000124  grad: 0.0743 (0.0858)  loss: 0.8451 (0.8463)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [9]  [4400/6250]  eta: 0:08:29  lr: 0.000124  grad: 0.0839 (0.0857)  loss: 0.8499 (0.8464)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [9]  [4500/6250]  eta: 0:08:01  lr: 0.000124  grad: 0.0756 (0.0856)  loss: 0.8449 (0.8464)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [9]  [4600/6250]  eta: 0:07:33  lr: 0.000124  grad: 0.0837 (0.0855)  loss: 0.8449 (0.8463)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [9]  [4700/6250]  eta: 0:07:06  lr: 0.000124  grad: 0.0777 (0.0854)  loss: 0.8498 (0.8463)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [9]  [4800/6250]  eta: 0:06:38  lr: 0.000124  grad: 0.0804 (0.0853)  loss: 0.8449 (0.8463)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [9]  [4900/6250]  eta: 0:06:10  lr: 0.000124  grad: 0.0783 (0.0853)  loss: 0.8446 (0.8463)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [9]  [5000/6250]  eta: 0:05:43  lr: 0.000124  grad: 0.0831 (0.0851)  loss: 0.8449 (0.8463)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [9]  [5100/6250]  eta: 0:05:15  lr: 0.000124  grad: 0.0730 (0.0850)  loss: 0.8468 (0.8463)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [9]  [5200/6250]  eta: 0:04:48  lr: 0.000124  grad: 0.0751 (0.0849)  loss: 0.8497 (0.8464)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [9]  [5300/6250]  eta: 0:04:20  lr: 0.000124  grad: 0.0749 (0.0848)  loss: 0.8496 (0.8464)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [9]  [5400/6250]  eta: 0:03:53  lr: 0.000124  grad: 0.0767 (0.0846)  loss: 0.8477 (0.8465)  time: 0.2748  data: 0.0002  max mem: 26157
Train: [9]  [5500/6250]  eta: 0:03:25  lr: 0.000124  grad: 0.0771 (0.0845)  loss: 0.8453 (0.8465)  time: 0.2772  data: 0.0003  max mem: 26157
Train: [9]  [5600/6250]  eta: 0:02:58  lr: 0.000124  grad: 0.0822 (0.0844)  loss: 0.8478 (0.8465)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [9]  [5700/6250]  eta: 0:02:31  lr: 0.000124  grad: 0.0759 (0.0843)  loss: 0.8441 (0.8465)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [9]  [5800/6250]  eta: 0:02:03  lr: 0.000124  grad: 0.0836 (0.0842)  loss: 0.8444 (0.8465)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [9]  [5900/6250]  eta: 0:01:36  lr: 0.000124  grad: 0.0726 (0.0841)  loss: 0.8543 (0.8465)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [9]  [6000/6250]  eta: 0:01:08  lr: 0.000124  grad: 0.0784 (0.0840)  loss: 0.8402 (0.8465)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [9]  [6100/6250]  eta: 0:00:41  lr: 0.000124  grad: 0.0823 (0.0840)  loss: 0.8481 (0.8465)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [9]  [6200/6250]  eta: 0:00:13  lr: 0.000124  grad: 0.0779 (0.0838)  loss: 0.8415 (0.8465)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [9]  [6249/6250]  eta: 0:00:00  lr: 0.000124  grad: 0.0783 (0.0838)  loss: 0.8489 (0.8465)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [9] Total time: 0:28:44 (0.2760 s / it)
Averaged stats: lr: 0.000124  grad: 0.0783 (0.0838)  loss: 0.8489 (0.8465)
Eval (hcp-train-subset): [9]  [ 0/62]  eta: 0:03:08  loss: 0.8631 (0.8631)  time: 3.0364  data: 2.9128  max mem: 26157
Eval (hcp-train-subset): [9]  [61/62]  eta: 0:00:00  loss: 0.8498 (0.8505)  time: 0.1325  data: 0.0481  max mem: 26157
Eval (hcp-train-subset): [9] Total time: 0:00:13 (0.2126 s / it)
Averaged stats (hcp-train-subset): loss: 0.8498 (0.8505)
Making plots (hcp-train-subset): example=49
Eval (hcp-val): [9]  [ 0/62]  eta: 0:05:21  loss: 0.8425 (0.8425)  time: 5.1816  data: 5.0973  max mem: 26157
Eval (hcp-val): [9]  [61/62]  eta: 0:00:00  loss: 0.8444 (0.8455)  time: 0.1326  data: 0.0478  max mem: 26157
Eval (hcp-val): [9] Total time: 0:00:12 (0.2033 s / it)
Averaged stats (hcp-val): loss: 0.8444 (0.8455)
Making plots (hcp-val): example=52
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [10]  [   0/6250]  eta: 10:59:21  lr: 0.000124  grad: 0.0744 (0.0744)  loss: 0.8397 (0.8397)  time: 6.3299  data: 6.0546  max mem: 26157
Train: [10]  [ 100/6250]  eta: 0:33:45  lr: 0.000124  grad: 0.0796 (0.0985)  loss: 0.8320 (0.8381)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [10]  [ 200/6250]  eta: 0:30:11  lr: 0.000124  grad: 0.0818 (0.0937)  loss: 0.8378 (0.8379)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [10]  [ 300/6250]  eta: 0:28:45  lr: 0.000124  grad: 0.0773 (0.0901)  loss: 0.8346 (0.8375)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [10]  [ 400/6250]  eta: 0:27:49  lr: 0.000124  grad: 0.0803 (0.0902)  loss: 0.8336 (0.8379)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [10]  [ 500/6250]  eta: 0:27:05  lr: 0.000124  grad: 0.0766 (0.0882)  loss: 0.8392 (0.8381)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [10]  [ 600/6250]  eta: 0:26:23  lr: 0.000124  grad: 0.0754 (0.0868)  loss: 0.8457 (0.8389)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [10]  [ 700/6250]  eta: 0:25:47  lr: 0.000124  grad: 0.0705 (0.0856)  loss: 0.8410 (0.8395)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [10]  [ 800/6250]  eta: 0:25:13  lr: 0.000124  grad: 0.0720 (0.0843)  loss: 0.8484 (0.8403)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [10]  [ 900/6250]  eta: 0:24:42  lr: 0.000124  grad: 0.0740 (0.0833)  loss: 0.8457 (0.8409)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [10]  [1000/6250]  eta: 0:24:11  lr: 0.000124  grad: 0.0708 (0.0823)  loss: 0.8454 (0.8415)  time: 0.2685  data: 0.0003  max mem: 26157
Train: [10]  [1100/6250]  eta: 0:23:40  lr: 0.000124  grad: 0.0739 (0.0817)  loss: 0.8432 (0.8418)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [10]  [1200/6250]  eta: 0:23:10  lr: 0.000124  grad: 0.0823 (0.0816)  loss: 0.8438 (0.8421)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [10]  [1300/6250]  eta: 0:22:40  lr: 0.000124  grad: 0.0722 (0.0813)  loss: 0.8455 (0.8423)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [10]  [1400/6250]  eta: 0:22:10  lr: 0.000124  grad: 0.0775 (0.0810)  loss: 0.8449 (0.8425)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [10]  [1500/6250]  eta: 0:21:41  lr: 0.000124  grad: 0.0735 (0.0807)  loss: 0.8452 (0.8426)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [10]  [1600/6250]  eta: 0:21:13  lr: 0.000124  grad: 0.0772 (0.0806)  loss: 0.8433 (0.8427)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [10]  [1700/6250]  eta: 0:20:48  lr: 0.000124  grad: 0.0790 (0.0804)  loss: 0.8421 (0.8428)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [10]  [1800/6250]  eta: 0:20:19  lr: 0.000124  grad: 0.0768 (0.0805)  loss: 0.8411 (0.8427)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [10]  [1900/6250]  eta: 0:19:51  lr: 0.000124  grad: 0.0721 (0.0806)  loss: 0.8444 (0.8428)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [10]  [2000/6250]  eta: 0:19:22  lr: 0.000124  grad: 0.0730 (0.0805)  loss: 0.8406 (0.8428)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [10]  [2100/6250]  eta: 0:18:54  lr: 0.000124  grad: 0.0720 (0.0803)  loss: 0.8459 (0.8429)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [10]  [2200/6250]  eta: 0:18:27  lr: 0.000124  grad: 0.0830 (0.0803)  loss: 0.8448 (0.8429)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [10]  [2300/6250]  eta: 0:17:59  lr: 0.000124  grad: 0.0750 (0.0802)  loss: 0.8383 (0.8428)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [10]  [2400/6250]  eta: 0:17:31  lr: 0.000124  grad: 0.0763 (0.0802)  loss: 0.8411 (0.8429)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [10]  [2500/6250]  eta: 0:17:03  lr: 0.000124  grad: 0.0785 (0.0802)  loss: 0.8436 (0.8429)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [10]  [2600/6250]  eta: 0:16:36  lr: 0.000124  grad: 0.0719 (0.0801)  loss: 0.8412 (0.8430)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [10]  [2700/6250]  eta: 0:16:15  lr: 0.000124  grad: 0.0731 (0.0800)  loss: 0.8446 (0.8430)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [10]  [2800/6250]  eta: 0:15:47  lr: 0.000124  grad: 0.0756 (0.0798)  loss: 0.8456 (0.8431)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [10]  [2900/6250]  eta: 0:15:19  lr: 0.000124  grad: 0.0796 (0.0798)  loss: 0.8398 (0.8430)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [10]  [3000/6250]  eta: 0:14:51  lr: 0.000124  grad: 0.0756 (0.0796)  loss: 0.8403 (0.8429)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [10]  [3100/6250]  eta: 0:14:23  lr: 0.000124  grad: 0.0763 (0.0796)  loss: 0.8404 (0.8429)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [10]  [3200/6250]  eta: 0:13:55  lr: 0.000124  grad: 0.0744 (0.0795)  loss: 0.8396 (0.8429)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [10]  [3300/6250]  eta: 0:13:27  lr: 0.000124  grad: 0.0745 (0.0796)  loss: 0.8356 (0.8428)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [10]  [3400/6250]  eta: 0:12:59  lr: 0.000124  grad: 0.0747 (0.0795)  loss: 0.8414 (0.8428)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [10]  [3500/6250]  eta: 0:12:32  lr: 0.000124  grad: 0.0773 (0.0794)  loss: 0.8425 (0.8428)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [10]  [3600/6250]  eta: 0:12:04  lr: 0.000124  grad: 0.0743 (0.0793)  loss: 0.8472 (0.8428)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [10]  [3700/6250]  eta: 0:11:36  lr: 0.000124  grad: 0.0722 (0.0793)  loss: 0.8414 (0.8428)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [10]  [3800/6250]  eta: 0:11:09  lr: 0.000124  grad: 0.0788 (0.0793)  loss: 0.8455 (0.8428)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [10]  [3900/6250]  eta: 0:10:41  lr: 0.000124  grad: 0.0718 (0.0792)  loss: 0.8436 (0.8428)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [10]  [4000/6250]  eta: 0:10:14  lr: 0.000124  grad: 0.0737 (0.0791)  loss: 0.8385 (0.8428)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [10]  [4100/6250]  eta: 0:09:46  lr: 0.000124  grad: 0.0756 (0.0791)  loss: 0.8455 (0.8428)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [10]  [4200/6250]  eta: 0:09:19  lr: 0.000124  grad: 0.0725 (0.0790)  loss: 0.8429 (0.8428)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [10]  [4300/6250]  eta: 0:08:51  lr: 0.000124  grad: 0.0735 (0.0789)  loss: 0.8451 (0.8428)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [10]  [4400/6250]  eta: 0:08:24  lr: 0.000124  grad: 0.0721 (0.0790)  loss: 0.8436 (0.8428)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [10]  [4500/6250]  eta: 0:07:57  lr: 0.000124  grad: 0.0738 (0.0789)  loss: 0.8433 (0.8428)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [10]  [4600/6250]  eta: 0:07:29  lr: 0.000124  grad: 0.0739 (0.0789)  loss: 0.8451 (0.8428)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [10]  [4700/6250]  eta: 0:07:02  lr: 0.000124  grad: 0.0725 (0.0787)  loss: 0.8415 (0.8428)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [10]  [4800/6250]  eta: 0:06:35  lr: 0.000124  grad: 0.0719 (0.0787)  loss: 0.8390 (0.8428)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [10]  [4900/6250]  eta: 0:06:07  lr: 0.000124  grad: 0.0726 (0.0786)  loss: 0.8436 (0.8427)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [10]  [5000/6250]  eta: 0:05:40  lr: 0.000124  grad: 0.0697 (0.0785)  loss: 0.8475 (0.8427)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [10]  [5100/6250]  eta: 0:05:13  lr: 0.000124  grad: 0.0742 (0.0784)  loss: 0.8414 (0.8427)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [10]  [5200/6250]  eta: 0:04:45  lr: 0.000124  grad: 0.0752 (0.0784)  loss: 0.8439 (0.8426)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [10]  [5300/6250]  eta: 0:04:18  lr: 0.000124  grad: 0.0774 (0.0784)  loss: 0.8416 (0.8426)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [10]  [5400/6250]  eta: 0:03:51  lr: 0.000124  grad: 0.0767 (0.0784)  loss: 0.8410 (0.8425)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [10]  [5500/6250]  eta: 0:03:24  lr: 0.000124  grad: 0.0753 (0.0783)  loss: 0.8362 (0.8425)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [10]  [5600/6250]  eta: 0:02:56  lr: 0.000124  grad: 0.0729 (0.0783)  loss: 0.8440 (0.8425)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [10]  [5700/6250]  eta: 0:02:29  lr: 0.000124  grad: 0.0683 (0.0782)  loss: 0.8457 (0.8425)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [10]  [5800/6250]  eta: 0:02:02  lr: 0.000124  grad: 0.0729 (0.0781)  loss: 0.8470 (0.8425)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [10]  [5900/6250]  eta: 0:01:35  lr: 0.000124  grad: 0.0720 (0.0781)  loss: 0.8416 (0.8424)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [10]  [6000/6250]  eta: 0:01:08  lr: 0.000124  grad: 0.0764 (0.0780)  loss: 0.8415 (0.8424)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [10]  [6100/6250]  eta: 0:00:40  lr: 0.000124  grad: 0.0710 (0.0780)  loss: 0.8424 (0.8424)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [10]  [6200/6250]  eta: 0:00:13  lr: 0.000124  grad: 0.0733 (0.0779)  loss: 0.8408 (0.8424)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [10]  [6249/6250]  eta: 0:00:00  lr: 0.000124  grad: 0.0749 (0.0779)  loss: 0.8438 (0.8424)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [10] Total time: 0:28:31 (0.2738 s / it)
Averaged stats: lr: 0.000124  grad: 0.0749 (0.0779)  loss: 0.8438 (0.8424)
Eval (hcp-train-subset): [10]  [ 0/62]  eta: 0:04:47  loss: 0.8566 (0.8566)  time: 4.6313  data: 4.5475  max mem: 26157
Eval (hcp-train-subset): [10]  [61/62]  eta: 0:00:00  loss: 0.8471 (0.8477)  time: 0.1332  data: 0.0487  max mem: 26157
Eval (hcp-train-subset): [10] Total time: 0:00:13 (0.2157 s / it)
Averaged stats (hcp-train-subset): loss: 0.8471 (0.8477)
Making plots (hcp-train-subset): example=58
Eval (hcp-val): [10]  [ 0/62]  eta: 0:03:54  loss: 0.8390 (0.8390)  time: 3.7797  data: 3.6541  max mem: 26157
Eval (hcp-val): [10]  [61/62]  eta: 0:00:00  loss: 0.8417 (0.8426)  time: 0.1297  data: 0.0452  max mem: 26157
Eval (hcp-val): [10] Total time: 0:00:14 (0.2368 s / it)
Averaged stats (hcp-val): loss: 0.8417 (0.8426)
Making plots (hcp-val): example=55
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [11]  [   0/6250]  eta: 10:57:05  lr: 0.000124  grad: 0.0578 (0.0578)  loss: 0.8646 (0.8646)  time: 6.3081  data: 5.9466  max mem: 26157
Train: [11]  [ 100/6250]  eta: 0:35:19  lr: 0.000124  grad: 0.0795 (0.0885)  loss: 0.8505 (0.8485)  time: 0.2734  data: 0.0003  max mem: 26157
Train: [11]  [ 200/6250]  eta: 0:31:04  lr: 0.000124  grad: 0.0760 (0.0855)  loss: 0.8417 (0.8458)  time: 0.2741  data: 0.0003  max mem: 26157
Train: [11]  [ 300/6250]  eta: 0:29:17  lr: 0.000124  grad: 0.0713 (0.0823)  loss: 0.8458 (0.8456)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [11]  [ 400/6250]  eta: 0:28:10  lr: 0.000124  grad: 0.0768 (0.0807)  loss: 0.8426 (0.8449)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [11]  [ 500/6250]  eta: 0:27:20  lr: 0.000124  grad: 0.0741 (0.0792)  loss: 0.8403 (0.8444)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [11]  [ 600/6250]  eta: 0:26:37  lr: 0.000124  grad: 0.0697 (0.0790)  loss: 0.8455 (0.8442)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [11]  [ 700/6250]  eta: 0:25:59  lr: 0.000124  grad: 0.0698 (0.0787)  loss: 0.8451 (0.8440)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [11]  [ 800/6250]  eta: 0:25:24  lr: 0.000124  grad: 0.0734 (0.0786)  loss: 0.8475 (0.8440)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [11]  [ 900/6250]  eta: 0:24:50  lr: 0.000124  grad: 0.0713 (0.0782)  loss: 0.8447 (0.8438)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [11]  [1000/6250]  eta: 0:24:18  lr: 0.000124  grad: 0.0694 (0.0777)  loss: 0.8448 (0.8440)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [11]  [1100/6250]  eta: 0:24:04  lr: 0.000124  grad: 0.0674 (0.0771)  loss: 0.8424 (0.8440)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [11]  [1200/6250]  eta: 0:23:56  lr: 0.000124  grad: 0.0701 (0.0767)  loss: 0.8517 (0.8441)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [11]  [1300/6250]  eta: 0:23:23  lr: 0.000124  grad: 0.0727 (0.0766)  loss: 0.8431 (0.8441)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [11]  [1400/6250]  eta: 0:22:49  lr: 0.000124  grad: 0.0734 (0.0763)  loss: 0.8467 (0.8441)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [11]  [1500/6250]  eta: 0:22:17  lr: 0.000124  grad: 0.0702 (0.0761)  loss: 0.8461 (0.8440)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [11]  [1600/6250]  eta: 0:21:45  lr: 0.000124  grad: 0.0708 (0.0758)  loss: 0.8386 (0.8439)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [11]  [1700/6250]  eta: 0:21:14  lr: 0.000124  grad: 0.0702 (0.0758)  loss: 0.8380 (0.8437)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [11]  [1800/6250]  eta: 0:20:43  lr: 0.000124  grad: 0.0731 (0.0758)  loss: 0.8402 (0.8436)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [11]  [1900/6250]  eta: 0:20:13  lr: 0.000124  grad: 0.0717 (0.0756)  loss: 0.8421 (0.8434)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [11]  [2000/6250]  eta: 0:19:43  lr: 0.000124  grad: 0.0721 (0.0754)  loss: 0.8413 (0.8433)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [11]  [2100/6250]  eta: 0:19:13  lr: 0.000124  grad: 0.0705 (0.0754)  loss: 0.8436 (0.8431)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [11]  [2200/6250]  eta: 0:18:44  lr: 0.000124  grad: 0.0729 (0.0754)  loss: 0.8363 (0.8429)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [11]  [2300/6250]  eta: 0:18:15  lr: 0.000124  grad: 0.0725 (0.0754)  loss: 0.8333 (0.8427)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [11]  [2400/6250]  eta: 0:17:46  lr: 0.000124  grad: 0.0722 (0.0753)  loss: 0.8385 (0.8425)  time: 0.2748  data: 0.0002  max mem: 26157
Train: [11]  [2500/6250]  eta: 0:17:30  lr: 0.000124  grad: 0.0697 (0.0753)  loss: 0.8384 (0.8423)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [11]  [2600/6250]  eta: 0:17:00  lr: 0.000124  grad: 0.0753 (0.0753)  loss: 0.8400 (0.8423)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [11]  [2700/6250]  eta: 0:16:31  lr: 0.000124  grad: 0.0708 (0.0752)  loss: 0.8409 (0.8422)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [11]  [2800/6250]  eta: 0:16:02  lr: 0.000124  grad: 0.0757 (0.0753)  loss: 0.8417 (0.8420)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [11]  [2900/6250]  eta: 0:15:33  lr: 0.000124  grad: 0.0765 (0.0754)  loss: 0.8331 (0.8419)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [11]  [3000/6250]  eta: 0:15:04  lr: 0.000124  grad: 0.0768 (0.0753)  loss: 0.8358 (0.8418)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [11]  [3100/6250]  eta: 0:14:35  lr: 0.000124  grad: 0.0725 (0.0753)  loss: 0.8413 (0.8417)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [11]  [3200/6250]  eta: 0:14:06  lr: 0.000124  grad: 0.0706 (0.0753)  loss: 0.8377 (0.8416)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [11]  [3300/6250]  eta: 0:13:38  lr: 0.000124  grad: 0.0685 (0.0753)  loss: 0.8405 (0.8415)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [11]  [3400/6250]  eta: 0:13:10  lr: 0.000124  grad: 0.0671 (0.0752)  loss: 0.8381 (0.8414)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [11]  [3500/6250]  eta: 0:12:41  lr: 0.000124  grad: 0.0715 (0.0752)  loss: 0.8394 (0.8414)  time: 0.2754  data: 0.0003  max mem: 26157
Train: [11]  [3600/6250]  eta: 0:12:13  lr: 0.000124  grad: 0.0686 (0.0751)  loss: 0.8409 (0.8413)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [11]  [3700/6250]  eta: 0:11:45  lr: 0.000124  grad: 0.0694 (0.0750)  loss: 0.8349 (0.8413)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [11]  [3800/6250]  eta: 0:11:17  lr: 0.000124  grad: 0.0740 (0.0750)  loss: 0.8371 (0.8413)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [11]  [3900/6250]  eta: 0:10:49  lr: 0.000124  grad: 0.0733 (0.0750)  loss: 0.8421 (0.8412)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [11]  [4000/6250]  eta: 0:10:21  lr: 0.000123  grad: 0.0706 (0.0750)  loss: 0.8389 (0.8412)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [11]  [4100/6250]  eta: 0:09:53  lr: 0.000123  grad: 0.0711 (0.0750)  loss: 0.8401 (0.8411)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [11]  [4200/6250]  eta: 0:09:25  lr: 0.000123  grad: 0.0714 (0.0749)  loss: 0.8368 (0.8411)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [11]  [4300/6250]  eta: 0:08:57  lr: 0.000123  grad: 0.0698 (0.0750)  loss: 0.8400 (0.8410)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [11]  [4400/6250]  eta: 0:08:29  lr: 0.000123  grad: 0.0707 (0.0750)  loss: 0.8396 (0.8409)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [11]  [4500/6250]  eta: 0:08:01  lr: 0.000123  grad: 0.0760 (0.0750)  loss: 0.8310 (0.8408)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [11]  [4600/6250]  eta: 0:07:34  lr: 0.000123  grad: 0.0733 (0.0750)  loss: 0.8322 (0.8408)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [11]  [4700/6250]  eta: 0:07:06  lr: 0.000123  grad: 0.0703 (0.0749)  loss: 0.8374 (0.8407)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [11]  [4800/6250]  eta: 0:06:38  lr: 0.000123  grad: 0.0731 (0.0749)  loss: 0.8387 (0.8406)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [11]  [4900/6250]  eta: 0:06:11  lr: 0.000123  grad: 0.0726 (0.0749)  loss: 0.8351 (0.8405)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [11]  [5000/6250]  eta: 0:05:43  lr: 0.000123  grad: 0.0738 (0.0749)  loss: 0.8359 (0.8404)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [11]  [5100/6250]  eta: 0:05:16  lr: 0.000123  grad: 0.0697 (0.0749)  loss: 0.8331 (0.8403)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [11]  [5200/6250]  eta: 0:04:48  lr: 0.000123  grad: 0.0707 (0.0749)  loss: 0.8340 (0.8402)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [11]  [5300/6250]  eta: 0:04:20  lr: 0.000123  grad: 0.0723 (0.0749)  loss: 0.8328 (0.8401)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [11]  [5400/6250]  eta: 0:03:53  lr: 0.000123  grad: 0.0755 (0.0749)  loss: 0.8368 (0.8401)  time: 0.3496  data: 0.0787  max mem: 26157
Train: [11]  [5500/6250]  eta: 0:03:26  lr: 0.000123  grad: 0.0772 (0.0749)  loss: 0.8396 (0.8400)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [11]  [5600/6250]  eta: 0:02:58  lr: 0.000123  grad: 0.0733 (0.0749)  loss: 0.8359 (0.8399)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [11]  [5700/6250]  eta: 0:02:31  lr: 0.000123  grad: 0.0724 (0.0749)  loss: 0.8351 (0.8399)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [11]  [5800/6250]  eta: 0:02:03  lr: 0.000123  grad: 0.0732 (0.0749)  loss: 0.8365 (0.8399)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [11]  [5900/6250]  eta: 0:01:36  lr: 0.000123  grad: 0.0709 (0.0748)  loss: 0.8397 (0.8398)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [11]  [6000/6250]  eta: 0:01:08  lr: 0.000123  grad: 0.0734 (0.0748)  loss: 0.8341 (0.8398)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [11]  [6100/6250]  eta: 0:00:41  lr: 0.000123  grad: 0.0735 (0.0748)  loss: 0.8358 (0.8398)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [11]  [6200/6250]  eta: 0:00:13  lr: 0.000123  grad: 0.0722 (0.0747)  loss: 0.8361 (0.8397)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [11]  [6249/6250]  eta: 0:00:00  lr: 0.000123  grad: 0.0715 (0.0747)  loss: 0.8368 (0.8397)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [11] Total time: 0:28:40 (0.2752 s / it)
Averaged stats: lr: 0.000123  grad: 0.0715 (0.0747)  loss: 0.8368 (0.8397)
Eval (hcp-train-subset): [11]  [ 0/62]  eta: 0:05:37  loss: 0.8532 (0.8532)  time: 5.4490  data: 5.3648  max mem: 26157
Eval (hcp-train-subset): [11]  [61/62]  eta: 0:00:00  loss: 0.8456 (0.8458)  time: 0.1118  data: 0.0291  max mem: 26157
Eval (hcp-train-subset): [11] Total time: 0:00:12 (0.2056 s / it)
Averaged stats (hcp-train-subset): loss: 0.8456 (0.8458)
Making plots (hcp-train-subset): example=41
Eval (hcp-val): [11]  [ 0/62]  eta: 0:04:19  loss: 0.8391 (0.8391)  time: 4.1866  data: 4.0725  max mem: 26157
Eval (hcp-val): [11]  [61/62]  eta: 0:00:00  loss: 0.8400 (0.8408)  time: 0.1236  data: 0.0406  max mem: 26157
Eval (hcp-val): [11] Total time: 0:00:13 (0.2216 s / it)
Averaged stats (hcp-val): loss: 0.8400 (0.8408)
Making plots (hcp-val): example=33
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [12]  [   0/6250]  eta: 7:02:41  lr: 0.000123  grad: 0.0528 (0.0528)  loss: 0.8708 (0.8708)  time: 4.0578  data: 3.6931  max mem: 26157
Train: [12]  [ 100/6250]  eta: 0:33:44  lr: 0.000123  grad: 0.0809 (0.0763)  loss: 0.8501 (0.8551)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [12]  [ 200/6250]  eta: 0:30:17  lr: 0.000123  grad: 0.0687 (0.0754)  loss: 0.8470 (0.8500)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [12]  [ 300/6250]  eta: 0:28:45  lr: 0.000123  grad: 0.0694 (0.0741)  loss: 0.8481 (0.8485)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [12]  [ 400/6250]  eta: 0:27:46  lr: 0.000123  grad: 0.0778 (0.0750)  loss: 0.8408 (0.8469)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [12]  [ 500/6250]  eta: 0:27:00  lr: 0.000123  grad: 0.0716 (0.0742)  loss: 0.8373 (0.8456)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [12]  [ 600/6250]  eta: 0:26:19  lr: 0.000123  grad: 0.0743 (0.0742)  loss: 0.8387 (0.8441)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [12]  [ 700/6250]  eta: 0:25:44  lr: 0.000123  grad: 0.0711 (0.0746)  loss: 0.8423 (0.8432)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [12]  [ 800/6250]  eta: 0:25:10  lr: 0.000123  grad: 0.0752 (0.0745)  loss: 0.8408 (0.8424)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [12]  [ 900/6250]  eta: 0:24:40  lr: 0.000123  grad: 0.0679 (0.0744)  loss: 0.8387 (0.8419)  time: 0.2699  data: 0.0003  max mem: 26157
Train: [12]  [1000/6250]  eta: 0:24:08  lr: 0.000123  grad: 0.0702 (0.0741)  loss: 0.8368 (0.8414)  time: 0.2709  data: 0.0003  max mem: 26157
Train: [12]  [1100/6250]  eta: 0:23:39  lr: 0.000123  grad: 0.0686 (0.0742)  loss: 0.8398 (0.8411)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [12]  [1200/6250]  eta: 0:23:09  lr: 0.000123  grad: 0.0678 (0.0740)  loss: 0.8373 (0.8407)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [12]  [1300/6250]  eta: 0:23:00  lr: 0.000123  grad: 0.0706 (0.0739)  loss: 0.8367 (0.8405)  time: 0.5363  data: 0.2641  max mem: 26157
Train: [12]  [1400/6250]  eta: 0:22:29  lr: 0.000123  grad: 0.0708 (0.0738)  loss: 0.8339 (0.8403)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [12]  [1500/6250]  eta: 0:21:59  lr: 0.000123  grad: 0.0716 (0.0738)  loss: 0.8384 (0.8402)  time: 0.2755  data: 0.0002  max mem: 26157
Train: [12]  [1600/6250]  eta: 0:21:29  lr: 0.000123  grad: 0.0670 (0.0736)  loss: 0.8393 (0.8400)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [12]  [1700/6250]  eta: 0:20:59  lr: 0.000123  grad: 0.0713 (0.0735)  loss: 0.8377 (0.8399)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [12]  [1800/6250]  eta: 0:20:29  lr: 0.000123  grad: 0.0713 (0.0735)  loss: 0.8378 (0.8398)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [12]  [1900/6250]  eta: 0:20:00  lr: 0.000123  grad: 0.0675 (0.0734)  loss: 0.8422 (0.8397)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [12]  [2000/6250]  eta: 0:19:31  lr: 0.000123  grad: 0.0694 (0.0733)  loss: 0.8411 (0.8397)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [12]  [2100/6250]  eta: 0:19:02  lr: 0.000123  grad: 0.0650 (0.0731)  loss: 0.8385 (0.8398)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [12]  [2200/6250]  eta: 0:18:34  lr: 0.000123  grad: 0.0694 (0.0730)  loss: 0.8361 (0.8399)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [12]  [2300/6250]  eta: 0:18:05  lr: 0.000123  grad: 0.0710 (0.0730)  loss: 0.8423 (0.8399)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [12]  [2400/6250]  eta: 0:17:37  lr: 0.000123  grad: 0.0687 (0.0729)  loss: 0.8395 (0.8398)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [12]  [2500/6250]  eta: 0:17:09  lr: 0.000123  grad: 0.0696 (0.0729)  loss: 0.8309 (0.8397)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [12]  [2600/6250]  eta: 0:16:40  lr: 0.000123  grad: 0.0719 (0.0729)  loss: 0.8374 (0.8397)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [12]  [2700/6250]  eta: 0:16:12  lr: 0.000123  grad: 0.0720 (0.0729)  loss: 0.8361 (0.8398)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [12]  [2800/6250]  eta: 0:15:44  lr: 0.000123  grad: 0.0709 (0.0728)  loss: 0.8426 (0.8398)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [12]  [2900/6250]  eta: 0:15:17  lr: 0.000123  grad: 0.0710 (0.0727)  loss: 0.8343 (0.8398)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [12]  [3000/6250]  eta: 0:14:49  lr: 0.000123  grad: 0.0703 (0.0727)  loss: 0.8369 (0.8397)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [12]  [3100/6250]  eta: 0:14:21  lr: 0.000123  grad: 0.0723 (0.0728)  loss: 0.8328 (0.8395)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [12]  [3200/6250]  eta: 0:13:53  lr: 0.000123  grad: 0.0704 (0.0728)  loss: 0.8387 (0.8395)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [12]  [3300/6250]  eta: 0:13:26  lr: 0.000123  grad: 0.0700 (0.0728)  loss: 0.8394 (0.8394)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [12]  [3400/6250]  eta: 0:12:58  lr: 0.000123  grad: 0.0687 (0.0729)  loss: 0.8369 (0.8393)  time: 0.2695  data: 0.0001  max mem: 26157
Train: [12]  [3500/6250]  eta: 0:12:31  lr: 0.000123  grad: 0.0689 (0.0729)  loss: 0.8351 (0.8392)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [12]  [3600/6250]  eta: 0:12:03  lr: 0.000123  grad: 0.0795 (0.0729)  loss: 0.8310 (0.8391)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [12]  [3700/6250]  eta: 0:11:38  lr: 0.000123  grad: 0.0707 (0.0728)  loss: 0.8363 (0.8390)  time: 0.4406  data: 0.1663  max mem: 26157
Train: [12]  [3800/6250]  eta: 0:11:10  lr: 0.000123  grad: 0.0736 (0.0729)  loss: 0.8368 (0.8389)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [12]  [3900/6250]  eta: 0:10:43  lr: 0.000123  grad: 0.0708 (0.0728)  loss: 0.8392 (0.8388)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [12]  [4000/6250]  eta: 0:10:20  lr: 0.000123  grad: 0.0686 (0.0728)  loss: 0.8342 (0.8388)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [12]  [4100/6250]  eta: 0:09:54  lr: 0.000123  grad: 0.0692 (0.0727)  loss: 0.8356 (0.8387)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [12]  [4200/6250]  eta: 0:09:26  lr: 0.000123  grad: 0.0739 (0.0727)  loss: 0.8372 (0.8386)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [12]  [4300/6250]  eta: 0:08:58  lr: 0.000123  grad: 0.0687 (0.0727)  loss: 0.8376 (0.8385)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [12]  [4400/6250]  eta: 0:08:30  lr: 0.000123  grad: 0.0706 (0.0727)  loss: 0.8352 (0.8385)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [12]  [4500/6250]  eta: 0:08:02  lr: 0.000123  grad: 0.0707 (0.0726)  loss: 0.8311 (0.8385)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [12]  [4600/6250]  eta: 0:07:35  lr: 0.000123  grad: 0.0678 (0.0727)  loss: 0.8394 (0.8384)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [12]  [4700/6250]  eta: 0:07:07  lr: 0.000123  grad: 0.0701 (0.0727)  loss: 0.8326 (0.8383)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [12]  [4800/6250]  eta: 0:06:39  lr: 0.000123  grad: 0.0705 (0.0727)  loss: 0.8349 (0.8382)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [12]  [4900/6250]  eta: 0:06:11  lr: 0.000123  grad: 0.0711 (0.0727)  loss: 0.8290 (0.8381)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [12]  [5000/6250]  eta: 0:05:44  lr: 0.000123  grad: 0.0718 (0.0727)  loss: 0.8303 (0.8381)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [12]  [5100/6250]  eta: 0:05:16  lr: 0.000123  grad: 0.0707 (0.0727)  loss: 0.8266 (0.8379)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [12]  [5200/6250]  eta: 0:04:48  lr: 0.000123  grad: 0.0751 (0.0727)  loss: 0.8323 (0.8378)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [12]  [5300/6250]  eta: 0:04:21  lr: 0.000123  grad: 0.0699 (0.0727)  loss: 0.8365 (0.8377)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [12]  [5400/6250]  eta: 0:03:53  lr: 0.000123  grad: 0.0704 (0.0727)  loss: 0.8319 (0.8377)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [12]  [5500/6250]  eta: 0:03:26  lr: 0.000123  grad: 0.0724 (0.0727)  loss: 0.8311 (0.8376)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [12]  [5600/6250]  eta: 0:02:58  lr: 0.000123  grad: 0.0712 (0.0727)  loss: 0.8329 (0.8375)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [12]  [5700/6250]  eta: 0:02:31  lr: 0.000123  grad: 0.0758 (0.0727)  loss: 0.8338 (0.8375)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [12]  [5800/6250]  eta: 0:02:03  lr: 0.000123  grad: 0.0703 (0.0727)  loss: 0.8325 (0.8373)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [12]  [5900/6250]  eta: 0:01:36  lr: 0.000123  grad: 0.0720 (0.0727)  loss: 0.8331 (0.8372)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [12]  [6000/6250]  eta: 0:01:08  lr: 0.000123  grad: 0.0758 (0.0727)  loss: 0.8302 (0.8371)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [12]  [6100/6250]  eta: 0:00:41  lr: 0.000123  grad: 0.0701 (0.0727)  loss: 0.8350 (0.8371)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [12]  [6200/6250]  eta: 0:00:13  lr: 0.000123  grad: 0.0702 (0.0727)  loss: 0.8304 (0.8370)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [12]  [6249/6250]  eta: 0:00:00  lr: 0.000123  grad: 0.0686 (0.0727)  loss: 0.8277 (0.8369)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [12] Total time: 0:28:38 (0.2750 s / it)
Averaged stats: lr: 0.000123  grad: 0.0686 (0.0727)  loss: 0.8277 (0.8369)
Eval (hcp-train-subset): [12]  [ 0/62]  eta: 0:05:58  loss: 0.8547 (0.8547)  time: 5.7887  data: 5.7046  max mem: 26157
Eval (hcp-train-subset): [12]  [61/62]  eta: 0:00:00  loss: 0.8446 (0.8441)  time: 0.1186  data: 0.0360  max mem: 26157
Eval (hcp-train-subset): [12] Total time: 0:00:13 (0.2145 s / it)
Averaged stats (hcp-train-subset): loss: 0.8446 (0.8441)
Making plots (hcp-train-subset): example=2
Eval (hcp-val): [12]  [ 0/62]  eta: 0:05:48  loss: 0.8329 (0.8329)  time: 5.6236  data: 5.5318  max mem: 26157
Eval (hcp-val): [12]  [61/62]  eta: 0:00:00  loss: 0.8365 (0.8386)  time: 0.1286  data: 0.0460  max mem: 26157
Eval (hcp-val): [12] Total time: 0:00:13 (0.2201 s / it)
Averaged stats (hcp-val): loss: 0.8365 (0.8386)
Making plots (hcp-val): example=25
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [13]  [   0/6250]  eta: 9:39:16  lr: 0.000123  grad: 0.1255 (0.1255)  loss: 0.7730 (0.7730)  time: 5.5610  data: 5.2304  max mem: 26157
Train: [13]  [ 100/6250]  eta: 0:33:55  lr: 0.000123  grad: 0.0721 (0.0859)  loss: 0.8452 (0.8412)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [13]  [ 200/6250]  eta: 0:30:19  lr: 0.000123  grad: 0.0735 (0.0826)  loss: 0.8388 (0.8403)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [13]  [ 300/6250]  eta: 0:28:49  lr: 0.000123  grad: 0.0763 (0.0810)  loss: 0.8361 (0.8394)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [13]  [ 400/6250]  eta: 0:27:52  lr: 0.000123  grad: 0.0707 (0.0798)  loss: 0.8382 (0.8387)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [13]  [ 500/6250]  eta: 0:27:09  lr: 0.000123  grad: 0.0677 (0.0784)  loss: 0.8337 (0.8377)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [13]  [ 600/6250]  eta: 0:26:30  lr: 0.000123  grad: 0.0712 (0.0774)  loss: 0.8379 (0.8372)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [13]  [ 700/6250]  eta: 0:25:55  lr: 0.000123  grad: 0.0665 (0.0765)  loss: 0.8337 (0.8366)  time: 0.2757  data: 0.0002  max mem: 26157
Train: [13]  [ 800/6250]  eta: 0:25:21  lr: 0.000123  grad: 0.0693 (0.0760)  loss: 0.8381 (0.8365)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [13]  [ 900/6250]  eta: 0:24:48  lr: 0.000123  grad: 0.0697 (0.0755)  loss: 0.8324 (0.8364)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [13]  [1000/6250]  eta: 0:24:16  lr: 0.000123  grad: 0.0691 (0.0749)  loss: 0.8317 (0.8362)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [13]  [1100/6250]  eta: 0:23:44  lr: 0.000123  grad: 0.0655 (0.0744)  loss: 0.8368 (0.8362)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [13]  [1200/6250]  eta: 0:23:13  lr: 0.000123  grad: 0.0662 (0.0740)  loss: 0.8364 (0.8362)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [13]  [1300/6250]  eta: 0:22:44  lr: 0.000123  grad: 0.0700 (0.0739)  loss: 0.8379 (0.8360)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [13]  [1400/6250]  eta: 0:22:14  lr: 0.000123  grad: 0.0655 (0.0734)  loss: 0.8411 (0.8361)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [13]  [1500/6250]  eta: 0:21:44  lr: 0.000123  grad: 0.0713 (0.0734)  loss: 0.8357 (0.8361)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [13]  [1600/6250]  eta: 0:21:15  lr: 0.000123  grad: 0.0645 (0.0731)  loss: 0.8412 (0.8362)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [13]  [1700/6250]  eta: 0:20:46  lr: 0.000123  grad: 0.0687 (0.0730)  loss: 0.8384 (0.8361)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [13]  [1800/6250]  eta: 0:20:18  lr: 0.000123  grad: 0.0710 (0.0729)  loss: 0.8389 (0.8362)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [13]  [1900/6250]  eta: 0:19:50  lr: 0.000123  grad: 0.0687 (0.0727)  loss: 0.8374 (0.8363)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [13]  [2000/6250]  eta: 0:19:21  lr: 0.000123  grad: 0.0664 (0.0725)  loss: 0.8382 (0.8364)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [13]  [2100/6250]  eta: 0:18:54  lr: 0.000123  grad: 0.0708 (0.0724)  loss: 0.8342 (0.8365)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [13]  [2200/6250]  eta: 0:18:26  lr: 0.000123  grad: 0.0676 (0.0723)  loss: 0.8373 (0.8366)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [13]  [2300/6250]  eta: 0:17:58  lr: 0.000123  grad: 0.0677 (0.0722)  loss: 0.8360 (0.8366)  time: 0.2684  data: 0.0003  max mem: 26157
Train: [13]  [2400/6250]  eta: 0:17:30  lr: 0.000123  grad: 0.0745 (0.0722)  loss: 0.8376 (0.8366)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [13]  [2500/6250]  eta: 0:17:03  lr: 0.000123  grad: 0.0700 (0.0721)  loss: 0.8339 (0.8366)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [13]  [2600/6250]  eta: 0:16:35  lr: 0.000123  grad: 0.0734 (0.0721)  loss: 0.8333 (0.8365)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [13]  [2700/6250]  eta: 0:16:07  lr: 0.000123  grad: 0.0751 (0.0721)  loss: 0.8312 (0.8364)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [13]  [2800/6250]  eta: 0:15:40  lr: 0.000123  grad: 0.0677 (0.0721)  loss: 0.8373 (0.8363)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [13]  [2900/6250]  eta: 0:15:12  lr: 0.000123  grad: 0.0708 (0.0721)  loss: 0.8287 (0.8363)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [13]  [3000/6250]  eta: 0:14:45  lr: 0.000123  grad: 0.0700 (0.0721)  loss: 0.8373 (0.8362)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [13]  [3100/6250]  eta: 0:14:17  lr: 0.000123  grad: 0.0698 (0.0721)  loss: 0.8321 (0.8362)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [13]  [3200/6250]  eta: 0:13:49  lr: 0.000123  grad: 0.0678 (0.0721)  loss: 0.8368 (0.8361)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [13]  [3300/6250]  eta: 0:13:22  lr: 0.000123  grad: 0.0691 (0.0720)  loss: 0.8326 (0.8360)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [13]  [3400/6250]  eta: 0:12:55  lr: 0.000123  grad: 0.0727 (0.0721)  loss: 0.8294 (0.8358)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [13]  [3500/6250]  eta: 0:12:28  lr: 0.000123  grad: 0.0733 (0.0721)  loss: 0.8298 (0.8357)  time: 0.2945  data: 0.0178  max mem: 26157
Train: [13]  [3600/6250]  eta: 0:12:00  lr: 0.000123  grad: 0.0745 (0.0722)  loss: 0.8365 (0.8356)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [13]  [3700/6250]  eta: 0:11:33  lr: 0.000122  grad: 0.0675 (0.0722)  loss: 0.8336 (0.8356)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [13]  [3800/6250]  eta: 0:11:06  lr: 0.000122  grad: 0.0724 (0.0722)  loss: 0.8326 (0.8356)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [13]  [3900/6250]  eta: 0:10:38  lr: 0.000122  grad: 0.0724 (0.0724)  loss: 0.8310 (0.8355)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [13]  [4000/6250]  eta: 0:10:11  lr: 0.000122  grad: 0.0708 (0.0724)  loss: 0.8327 (0.8354)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [13]  [4100/6250]  eta: 0:09:44  lr: 0.000122  grad: 0.0735 (0.0724)  loss: 0.8341 (0.8354)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [13]  [4200/6250]  eta: 0:09:16  lr: 0.000122  grad: 0.0701 (0.0724)  loss: 0.8361 (0.8353)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [13]  [4300/6250]  eta: 0:08:49  lr: 0.000122  grad: 0.0756 (0.0725)  loss: 0.8313 (0.8353)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [13]  [4400/6250]  eta: 0:08:22  lr: 0.000122  grad: 0.0737 (0.0724)  loss: 0.8277 (0.8352)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [13]  [4500/6250]  eta: 0:07:55  lr: 0.000122  grad: 0.0709 (0.0725)  loss: 0.8294 (0.8351)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [13]  [4600/6250]  eta: 0:07:27  lr: 0.000122  grad: 0.0740 (0.0725)  loss: 0.8391 (0.8350)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [13]  [4700/6250]  eta: 0:07:00  lr: 0.000122  grad: 0.0719 (0.0725)  loss: 0.8317 (0.8350)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [13]  [4800/6250]  eta: 0:06:33  lr: 0.000122  grad: 0.0731 (0.0725)  loss: 0.8334 (0.8349)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [13]  [4900/6250]  eta: 0:06:06  lr: 0.000122  grad: 0.0712 (0.0726)  loss: 0.8291 (0.8348)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [13]  [5000/6250]  eta: 0:05:39  lr: 0.000122  grad: 0.0740 (0.0727)  loss: 0.8321 (0.8347)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [13]  [5100/6250]  eta: 0:05:11  lr: 0.000122  grad: 0.0725 (0.0727)  loss: 0.8308 (0.8347)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [13]  [5200/6250]  eta: 0:04:44  lr: 0.000122  grad: 0.0687 (0.0727)  loss: 0.8317 (0.8346)  time: 0.2778  data: 0.0003  max mem: 26157
Train: [13]  [5300/6250]  eta: 0:04:17  lr: 0.000122  grad: 0.0711 (0.0727)  loss: 0.8279 (0.8346)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [13]  [5400/6250]  eta: 0:03:50  lr: 0.000122  grad: 0.0708 (0.0727)  loss: 0.8306 (0.8345)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [13]  [5500/6250]  eta: 0:03:23  lr: 0.000122  grad: 0.0701 (0.0728)  loss: 0.8329 (0.8345)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [13]  [5600/6250]  eta: 0:02:56  lr: 0.000122  grad: 0.0705 (0.0729)  loss: 0.8325 (0.8345)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [13]  [5700/6250]  eta: 0:02:29  lr: 0.000122  grad: 0.0721 (0.0729)  loss: 0.8287 (0.8344)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [13]  [5800/6250]  eta: 0:02:02  lr: 0.000122  grad: 0.0738 (0.0730)  loss: 0.8253 (0.8343)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [13]  [5900/6250]  eta: 0:01:34  lr: 0.000122  grad: 0.0737 (0.0730)  loss: 0.8310 (0.8342)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [13]  [6000/6250]  eta: 0:01:07  lr: 0.000122  grad: 0.0701 (0.0731)  loss: 0.8319 (0.8341)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [13]  [6100/6250]  eta: 0:00:40  lr: 0.000122  grad: 0.0765 (0.0731)  loss: 0.8302 (0.8341)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [13]  [6200/6250]  eta: 0:00:13  lr: 0.000122  grad: 0.0701 (0.0731)  loss: 0.8382 (0.8341)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [13]  [6249/6250]  eta: 0:00:00  lr: 0.000122  grad: 0.0682 (0.0731)  loss: 0.8338 (0.8341)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [13] Total time: 0:28:19 (0.2719 s / it)
Averaged stats: lr: 0.000122  grad: 0.0682 (0.0731)  loss: 0.8338 (0.8341)
Eval (hcp-train-subset): [13]  [ 0/62]  eta: 0:05:14  loss: 0.8543 (0.8543)  time: 5.0772  data: 4.9930  max mem: 26157
Eval (hcp-train-subset): [13]  [61/62]  eta: 0:00:00  loss: 0.8418 (0.8427)  time: 0.1244  data: 0.0399  max mem: 26157
Eval (hcp-train-subset): [13] Total time: 0:00:12 (0.2047 s / it)
Averaged stats (hcp-train-subset): loss: 0.8418 (0.8427)
Making plots (hcp-train-subset): example=61
Eval (hcp-val): [13]  [ 0/62]  eta: 0:03:10  loss: 0.8306 (0.8306)  time: 3.0745  data: 2.9475  max mem: 26157
Eval (hcp-val): [13]  [61/62]  eta: 0:00:00  loss: 0.8376 (0.8377)  time: 0.1402  data: 0.0572  max mem: 26157
Eval (hcp-val): [13] Total time: 0:00:12 (0.2088 s / it)
Averaged stats (hcp-val): loss: 0.8376 (0.8377)
Making plots (hcp-val): example=36
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [14]  [   0/6250]  eta: 11:17:42  lr: 0.000122  grad: 0.0604 (0.0604)  loss: 0.8633 (0.8633)  time: 6.5060  data: 6.2310  max mem: 26157
Train: [14]  [ 100/6250]  eta: 0:35:01  lr: 0.000122  grad: 0.0761 (0.0848)  loss: 0.8379 (0.8407)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [14]  [ 200/6250]  eta: 0:30:54  lr: 0.000122  grad: 0.0729 (0.0826)  loss: 0.8293 (0.8371)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [14]  [ 300/6250]  eta: 0:29:11  lr: 0.000122  grad: 0.0684 (0.0793)  loss: 0.8355 (0.8357)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [14]  [ 400/6250]  eta: 0:28:05  lr: 0.000122  grad: 0.0691 (0.0774)  loss: 0.8351 (0.8356)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [14]  [ 500/6250]  eta: 0:27:16  lr: 0.000122  grad: 0.0693 (0.0762)  loss: 0.8308 (0.8360)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [14]  [ 600/6250]  eta: 0:26:35  lr: 0.000122  grad: 0.0697 (0.0754)  loss: 0.8317 (0.8358)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [14]  [ 700/6250]  eta: 0:25:57  lr: 0.000122  grad: 0.0711 (0.0754)  loss: 0.8297 (0.8353)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [14]  [ 800/6250]  eta: 0:25:21  lr: 0.000122  grad: 0.0694 (0.0751)  loss: 0.8341 (0.8352)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [14]  [ 900/6250]  eta: 0:24:49  lr: 0.000122  grad: 0.0688 (0.0747)  loss: 0.8366 (0.8351)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [14]  [1000/6250]  eta: 0:24:17  lr: 0.000122  grad: 0.0730 (0.0743)  loss: 0.8349 (0.8352)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [14]  [1100/6250]  eta: 0:23:46  lr: 0.000122  grad: 0.0714 (0.0740)  loss: 0.8361 (0.8353)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [14]  [1200/6250]  eta: 0:23:15  lr: 0.000122  grad: 0.0700 (0.0737)  loss: 0.8369 (0.8353)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [14]  [1300/6250]  eta: 0:22:45  lr: 0.000122  grad: 0.0712 (0.0734)  loss: 0.8457 (0.8354)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [14]  [1400/6250]  eta: 0:22:15  lr: 0.000122  grad: 0.0690 (0.0731)  loss: 0.8384 (0.8355)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [14]  [1500/6250]  eta: 0:21:46  lr: 0.000122  grad: 0.0704 (0.0729)  loss: 0.8359 (0.8355)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [14]  [1600/6250]  eta: 0:21:17  lr: 0.000122  grad: 0.0706 (0.0727)  loss: 0.8387 (0.8355)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [14]  [1700/6250]  eta: 0:20:49  lr: 0.000122  grad: 0.0680 (0.0726)  loss: 0.8346 (0.8356)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [14]  [1800/6250]  eta: 0:20:20  lr: 0.000122  grad: 0.0727 (0.0724)  loss: 0.8362 (0.8356)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [14]  [1900/6250]  eta: 0:19:51  lr: 0.000122  grad: 0.0693 (0.0723)  loss: 0.8378 (0.8356)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [14]  [2000/6250]  eta: 0:19:23  lr: 0.000122  grad: 0.0707 (0.0722)  loss: 0.8363 (0.8356)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [14]  [2100/6250]  eta: 0:18:55  lr: 0.000122  grad: 0.0668 (0.0722)  loss: 0.8376 (0.8355)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [14]  [2200/6250]  eta: 0:18:27  lr: 0.000122  grad: 0.0763 (0.0723)  loss: 0.8352 (0.8355)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [14]  [2300/6250]  eta: 0:17:58  lr: 0.000122  grad: 0.0693 (0.0722)  loss: 0.8340 (0.8355)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [14]  [2400/6250]  eta: 0:17:30  lr: 0.000122  grad: 0.0666 (0.0722)  loss: 0.8355 (0.8354)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [14]  [2500/6250]  eta: 0:17:03  lr: 0.000122  grad: 0.0678 (0.0722)  loss: 0.8301 (0.8353)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [14]  [2600/6250]  eta: 0:16:35  lr: 0.000122  grad: 0.0723 (0.0723)  loss: 0.8323 (0.8353)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [14]  [2700/6250]  eta: 0:16:07  lr: 0.000122  grad: 0.0671 (0.0722)  loss: 0.8361 (0.8353)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [14]  [2800/6250]  eta: 0:15:40  lr: 0.000122  grad: 0.0670 (0.0722)  loss: 0.8392 (0.8354)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [14]  [2900/6250]  eta: 0:15:12  lr: 0.000122  grad: 0.0687 (0.0721)  loss: 0.8346 (0.8353)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [14]  [3000/6250]  eta: 0:14:46  lr: 0.000122  grad: 0.0711 (0.0722)  loss: 0.8338 (0.8353)  time: 0.3347  data: 0.0590  max mem: 26157
Train: [14]  [3100/6250]  eta: 0:14:19  lr: 0.000122  grad: 0.0687 (0.0721)  loss: 0.8324 (0.8352)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [14]  [3200/6250]  eta: 0:13:58  lr: 0.000122  grad: 0.0731 (0.0721)  loss: 0.8312 (0.8351)  time: 0.2708  data: 0.0001  max mem: 26157
Train: [14]  [3300/6250]  eta: 0:13:30  lr: 0.000122  grad: 0.0692 (0.0721)  loss: 0.8368 (0.8350)  time: 0.2736  data: 0.0002  max mem: 26157
Train: [14]  [3400/6250]  eta: 0:13:06  lr: 0.000122  grad: 0.0704 (0.0721)  loss: 0.8264 (0.8349)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [14]  [3500/6250]  eta: 0:12:38  lr: 0.000122  grad: 0.0663 (0.0721)  loss: 0.8308 (0.8348)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [14]  [3600/6250]  eta: 0:12:10  lr: 0.000122  grad: 0.0700 (0.0720)  loss: 0.8325 (0.8347)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [14]  [3700/6250]  eta: 0:11:42  lr: 0.000122  grad: 0.0710 (0.0720)  loss: 0.8346 (0.8346)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [14]  [3800/6250]  eta: 0:11:14  lr: 0.000122  grad: 0.0704 (0.0720)  loss: 0.8319 (0.8345)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [14]  [3900/6250]  eta: 0:10:47  lr: 0.000122  grad: 0.0694 (0.0720)  loss: 0.8331 (0.8345)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [14]  [4000/6250]  eta: 0:10:19  lr: 0.000122  grad: 0.0695 (0.0720)  loss: 0.8293 (0.8345)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [14]  [4100/6250]  eta: 0:09:51  lr: 0.000122  grad: 0.0660 (0.0720)  loss: 0.8412 (0.8345)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [14]  [4200/6250]  eta: 0:09:23  lr: 0.000122  grad: 0.0697 (0.0720)  loss: 0.8415 (0.8345)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [14]  [4300/6250]  eta: 0:08:55  lr: 0.000122  grad: 0.0708 (0.0720)  loss: 0.8308 (0.8345)  time: 0.2711  data: 0.0003  max mem: 26157
Train: [14]  [4400/6250]  eta: 0:08:28  lr: 0.000122  grad: 0.0718 (0.0720)  loss: 0.8320 (0.8345)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [14]  [4500/6250]  eta: 0:08:00  lr: 0.000122  grad: 0.0684 (0.0719)  loss: 0.8340 (0.8344)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [14]  [4600/6250]  eta: 0:07:32  lr: 0.000122  grad: 0.0679 (0.0719)  loss: 0.8314 (0.8344)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [14]  [4700/6250]  eta: 0:07:05  lr: 0.000122  grad: 0.0687 (0.0719)  loss: 0.8318 (0.8343)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [14]  [4800/6250]  eta: 0:06:37  lr: 0.000122  grad: 0.0683 (0.0719)  loss: 0.8294 (0.8343)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [14]  [4900/6250]  eta: 0:06:10  lr: 0.000122  grad: 0.0674 (0.0719)  loss: 0.8295 (0.8342)  time: 0.2699  data: 0.0001  max mem: 26157
Train: [14]  [5000/6250]  eta: 0:05:42  lr: 0.000122  grad: 0.0684 (0.0719)  loss: 0.8301 (0.8342)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [14]  [5100/6250]  eta: 0:05:14  lr: 0.000122  grad: 0.0727 (0.0718)  loss: 0.8325 (0.8342)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [14]  [5200/6250]  eta: 0:04:47  lr: 0.000122  grad: 0.0699 (0.0718)  loss: 0.8371 (0.8342)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [14]  [5300/6250]  eta: 0:04:20  lr: 0.000122  grad: 0.0715 (0.0717)  loss: 0.8349 (0.8342)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [14]  [5400/6250]  eta: 0:03:52  lr: 0.000122  grad: 0.0669 (0.0718)  loss: 0.8343 (0.8342)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [14]  [5500/6250]  eta: 0:03:25  lr: 0.000122  grad: 0.0676 (0.0718)  loss: 0.8365 (0.8342)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [14]  [5600/6250]  eta: 0:02:57  lr: 0.000122  grad: 0.0759 (0.0718)  loss: 0.8314 (0.8342)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [14]  [5700/6250]  eta: 0:02:30  lr: 0.000122  grad: 0.0667 (0.0718)  loss: 0.8369 (0.8342)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [14]  [5800/6250]  eta: 0:02:03  lr: 0.000122  grad: 0.0673 (0.0718)  loss: 0.8329 (0.8342)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [14]  [5900/6250]  eta: 0:01:35  lr: 0.000122  grad: 0.0679 (0.0717)  loss: 0.8355 (0.8343)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [14]  [6000/6250]  eta: 0:01:08  lr: 0.000122  grad: 0.0702 (0.0717)  loss: 0.8346 (0.8343)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [14]  [6100/6250]  eta: 0:00:41  lr: 0.000122  grad: 0.0691 (0.0717)  loss: 0.8362 (0.8343)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [14]  [6200/6250]  eta: 0:00:13  lr: 0.000122  grad: 0.0689 (0.0718)  loss: 0.8293 (0.8342)  time: 0.2698  data: 0.0001  max mem: 26157
Train: [14]  [6249/6250]  eta: 0:00:00  lr: 0.000122  grad: 0.0690 (0.0718)  loss: 0.8328 (0.8342)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [14] Total time: 0:28:33 (0.2742 s / it)
Averaged stats: lr: 0.000122  grad: 0.0690 (0.0718)  loss: 0.8328 (0.8342)
Eval (hcp-train-subset): [14]  [ 0/62]  eta: 0:03:14  loss: 0.8519 (0.8519)  time: 3.1363  data: 3.0155  max mem: 26157
Eval (hcp-train-subset): [14]  [61/62]  eta: 0:00:00  loss: 0.8425 (0.8429)  time: 0.1272  data: 0.0437  max mem: 26157
Eval (hcp-train-subset): [14] Total time: 0:00:13 (0.2108 s / it)
Averaged stats (hcp-train-subset): loss: 0.8425 (0.8429)
Making plots (hcp-train-subset): example=40
Eval (hcp-val): [14]  [ 0/62]  eta: 0:05:43  loss: 0.8318 (0.8318)  time: 5.5418  data: 5.4582  max mem: 26157
Eval (hcp-val): [14]  [61/62]  eta: 0:00:00  loss: 0.8378 (0.8379)  time: 0.1292  data: 0.0445  max mem: 26157
Eval (hcp-val): [14] Total time: 0:00:12 (0.2084 s / it)
Averaged stats (hcp-val): loss: 0.8378 (0.8379)
Making plots (hcp-val): example=51
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [15]  [   0/6250]  eta: 8:26:10  lr: 0.000122  grad: 0.2194 (0.2194)  loss: 0.7838 (0.7838)  time: 4.8593  data: 4.5079  max mem: 26157
Train: [15]  [ 100/6250]  eta: 0:34:14  lr: 0.000122  grad: 0.0820 (0.0887)  loss: 0.8124 (0.8322)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [15]  [ 200/6250]  eta: 0:30:33  lr: 0.000122  grad: 0.0690 (0.0811)  loss: 0.8460 (0.8324)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [15]  [ 300/6250]  eta: 0:28:56  lr: 0.000122  grad: 0.0688 (0.0787)  loss: 0.8308 (0.8318)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [15]  [ 400/6250]  eta: 0:27:56  lr: 0.000122  grad: 0.0695 (0.0775)  loss: 0.8264 (0.8316)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [15]  [ 500/6250]  eta: 0:27:07  lr: 0.000122  grad: 0.0735 (0.0768)  loss: 0.8289 (0.8310)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [15]  [ 600/6250]  eta: 0:26:31  lr: 0.000122  grad: 0.0671 (0.0758)  loss: 0.8306 (0.8310)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [15]  [ 700/6250]  eta: 0:26:07  lr: 0.000122  grad: 0.0647 (0.0751)  loss: 0.8318 (0.8311)  time: 0.2719  data: 0.0006  max mem: 26157
Train: [15]  [ 800/6250]  eta: 0:25:41  lr: 0.000122  grad: 0.0657 (0.0741)  loss: 0.8328 (0.8315)  time: 0.2717  data: 0.0003  max mem: 26157
Train: [15]  [ 900/6250]  eta: 0:25:05  lr: 0.000122  grad: 0.0653 (0.0735)  loss: 0.8354 (0.8320)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [15]  [1000/6250]  eta: 0:24:45  lr: 0.000122  grad: 0.0742 (0.0731)  loss: 0.8308 (0.8322)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [15]  [1100/6250]  eta: 0:24:11  lr: 0.000121  grad: 0.0698 (0.0728)  loss: 0.8355 (0.8324)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [15]  [1200/6250]  eta: 0:23:45  lr: 0.000121  grad: 0.0683 (0.0726)  loss: 0.8326 (0.8325)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [15]  [1300/6250]  eta: 0:23:12  lr: 0.000121  grad: 0.0675 (0.0725)  loss: 0.8338 (0.8325)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [15]  [1400/6250]  eta: 0:22:39  lr: 0.000121  grad: 0.0662 (0.0724)  loss: 0.8355 (0.8325)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [15]  [1500/6250]  eta: 0:22:08  lr: 0.000121  grad: 0.0687 (0.0723)  loss: 0.8338 (0.8325)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [15]  [1600/6250]  eta: 0:21:37  lr: 0.000121  grad: 0.0656 (0.0724)  loss: 0.8359 (0.8325)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [15]  [1700/6250]  eta: 0:21:06  lr: 0.000121  grad: 0.0653 (0.0722)  loss: 0.8361 (0.8326)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [15]  [1800/6250]  eta: 0:20:36  lr: 0.000121  grad: 0.0716 (0.0723)  loss: 0.8293 (0.8325)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [15]  [1900/6250]  eta: 0:20:06  lr: 0.000121  grad: 0.0689 (0.0723)  loss: 0.8337 (0.8324)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [15]  [2000/6250]  eta: 0:19:36  lr: 0.000121  grad: 0.0697 (0.0723)  loss: 0.8328 (0.8324)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [15]  [2100/6250]  eta: 0:19:07  lr: 0.000121  grad: 0.0740 (0.0724)  loss: 0.8320 (0.8324)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [15]  [2200/6250]  eta: 0:18:38  lr: 0.000121  grad: 0.0681 (0.0724)  loss: 0.8273 (0.8323)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [15]  [2300/6250]  eta: 0:18:09  lr: 0.000121  grad: 0.0707 (0.0723)  loss: 0.8337 (0.8323)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [15]  [2400/6250]  eta: 0:17:41  lr: 0.000121  grad: 0.0669 (0.0723)  loss: 0.8346 (0.8323)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [15]  [2500/6250]  eta: 0:17:12  lr: 0.000121  grad: 0.0708 (0.0725)  loss: 0.8350 (0.8323)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [15]  [2600/6250]  eta: 0:16:44  lr: 0.000121  grad: 0.0722 (0.0726)  loss: 0.8343 (0.8322)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [15]  [2700/6250]  eta: 0:16:16  lr: 0.000121  grad: 0.0691 (0.0726)  loss: 0.8264 (0.8322)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [15]  [2800/6250]  eta: 0:15:48  lr: 0.000121  grad: 0.0685 (0.0727)  loss: 0.8299 (0.8321)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [15]  [2900/6250]  eta: 0:15:20  lr: 0.000121  grad: 0.0702 (0.0727)  loss: 0.8281 (0.8320)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [15]  [3000/6250]  eta: 0:14:52  lr: 0.000121  grad: 0.0677 (0.0728)  loss: 0.8287 (0.8319)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [15]  [3100/6250]  eta: 0:14:24  lr: 0.000121  grad: 0.0680 (0.0728)  loss: 0.8298 (0.8318)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [15]  [3200/6250]  eta: 0:13:56  lr: 0.000121  grad: 0.0747 (0.0728)  loss: 0.8292 (0.8318)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [15]  [3300/6250]  eta: 0:13:28  lr: 0.000121  grad: 0.0712 (0.0728)  loss: 0.8360 (0.8318)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [15]  [3400/6250]  eta: 0:13:01  lr: 0.000121  grad: 0.0673 (0.0728)  loss: 0.8354 (0.8319)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [15]  [3500/6250]  eta: 0:12:33  lr: 0.000121  grad: 0.0661 (0.0728)  loss: 0.8339 (0.8319)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [15]  [3600/6250]  eta: 0:12:05  lr: 0.000121  grad: 0.0702 (0.0727)  loss: 0.8306 (0.8318)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [15]  [3700/6250]  eta: 0:11:37  lr: 0.000121  grad: 0.0675 (0.0727)  loss: 0.8296 (0.8318)  time: 0.2701  data: 0.0001  max mem: 26157
Train: [15]  [3800/6250]  eta: 0:11:10  lr: 0.000121  grad: 0.0644 (0.0726)  loss: 0.8379 (0.8319)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [15]  [3900/6250]  eta: 0:10:42  lr: 0.000121  grad: 0.0704 (0.0726)  loss: 0.8298 (0.8319)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [15]  [4000/6250]  eta: 0:10:15  lr: 0.000121  grad: 0.0710 (0.0726)  loss: 0.8339 (0.8318)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [15]  [4100/6250]  eta: 0:09:47  lr: 0.000121  grad: 0.0711 (0.0726)  loss: 0.8362 (0.8319)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [15]  [4200/6250]  eta: 0:09:19  lr: 0.000121  grad: 0.0715 (0.0726)  loss: 0.8341 (0.8319)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [15]  [4300/6250]  eta: 0:08:52  lr: 0.000121  grad: 0.0686 (0.0726)  loss: 0.8323 (0.8319)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [15]  [4400/6250]  eta: 0:08:24  lr: 0.000121  grad: 0.0709 (0.0726)  loss: 0.8258 (0.8318)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [15]  [4500/6250]  eta: 0:07:57  lr: 0.000121  grad: 0.0720 (0.0726)  loss: 0.8301 (0.8318)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [15]  [4600/6250]  eta: 0:07:30  lr: 0.000121  grad: 0.0686 (0.0726)  loss: 0.8306 (0.8317)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [15]  [4700/6250]  eta: 0:07:03  lr: 0.000121  grad: 0.0750 (0.0726)  loss: 0.8285 (0.8317)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [15]  [4800/6250]  eta: 0:06:35  lr: 0.000121  grad: 0.0734 (0.0726)  loss: 0.8309 (0.8316)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [15]  [4900/6250]  eta: 0:06:08  lr: 0.000121  grad: 0.0688 (0.0726)  loss: 0.8355 (0.8316)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [15]  [5000/6250]  eta: 0:05:40  lr: 0.000121  grad: 0.0679 (0.0726)  loss: 0.8344 (0.8316)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [15]  [5100/6250]  eta: 0:05:13  lr: 0.000121  grad: 0.0671 (0.0726)  loss: 0.8355 (0.8316)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [15]  [5200/6250]  eta: 0:04:46  lr: 0.000121  grad: 0.0715 (0.0726)  loss: 0.8249 (0.8315)  time: 0.2760  data: 0.0002  max mem: 26157
Train: [15]  [5300/6250]  eta: 0:04:19  lr: 0.000121  grad: 0.0720 (0.0726)  loss: 0.8329 (0.8316)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [15]  [5400/6250]  eta: 0:03:52  lr: 0.000121  grad: 0.0701 (0.0726)  loss: 0.8348 (0.8316)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [15]  [5500/6250]  eta: 0:03:25  lr: 0.000121  grad: 0.0669 (0.0725)  loss: 0.8370 (0.8316)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [15]  [5600/6250]  eta: 0:02:57  lr: 0.000121  grad: 0.0730 (0.0725)  loss: 0.8320 (0.8316)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [15]  [5700/6250]  eta: 0:02:30  lr: 0.000121  grad: 0.0733 (0.0725)  loss: 0.8311 (0.8316)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [15]  [5800/6250]  eta: 0:02:02  lr: 0.000121  grad: 0.0671 (0.0725)  loss: 0.8324 (0.8316)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [15]  [5900/6250]  eta: 0:01:35  lr: 0.000121  grad: 0.0733 (0.0725)  loss: 0.8339 (0.8316)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [15]  [6000/6250]  eta: 0:01:08  lr: 0.000121  grad: 0.0672 (0.0725)  loss: 0.8318 (0.8316)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [15]  [6100/6250]  eta: 0:00:40  lr: 0.000121  grad: 0.0712 (0.0726)  loss: 0.8279 (0.8316)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [15]  [6200/6250]  eta: 0:00:13  lr: 0.000121  grad: 0.0669 (0.0728)  loss: 0.8317 (0.8316)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [15]  [6249/6250]  eta: 0:00:00  lr: 0.000121  grad: 0.0730 (0.0728)  loss: 0.8327 (0.8317)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [15] Total time: 0:28:37 (0.2749 s / it)
Averaged stats: lr: 0.000121  grad: 0.0730 (0.0728)  loss: 0.8327 (0.8317)
Eval (hcp-train-subset): [15]  [ 0/62]  eta: 0:03:18  loss: 0.8497 (0.8497)  time: 3.1989  data: 3.0566  max mem: 26157
Eval (hcp-train-subset): [15]  [61/62]  eta: 0:00:00  loss: 0.8382 (0.8394)  time: 0.1307  data: 0.0464  max mem: 26157
Eval (hcp-train-subset): [15] Total time: 0:00:12 (0.2082 s / it)
Averaged stats (hcp-train-subset): loss: 0.8382 (0.8394)
Making plots (hcp-train-subset): example=56
Eval (hcp-val): [15]  [ 0/62]  eta: 0:06:02  loss: 0.8309 (0.8309)  time: 5.8393  data: 5.7555  max mem: 26157
Eval (hcp-val): [15]  [61/62]  eta: 0:00:00  loss: 0.8349 (0.8355)  time: 0.1184  data: 0.0357  max mem: 26157
Eval (hcp-val): [15] Total time: 0:00:12 (0.2059 s / it)
Averaged stats (hcp-val): loss: 0.8349 (0.8355)
Making plots (hcp-val): example=42
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [16]  [   0/6250]  eta: 10:29:25  lr: 0.000121  grad: 0.0766 (0.0766)  loss: 0.8514 (0.8514)  time: 6.0425  data: 5.7599  max mem: 26157
Train: [16]  [ 100/6250]  eta: 0:33:40  lr: 0.000121  grad: 0.0770 (0.0819)  loss: 0.8296 (0.8347)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [16]  [ 200/6250]  eta: 0:30:12  lr: 0.000121  grad: 0.0727 (0.0784)  loss: 0.8290 (0.8324)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [16]  [ 300/6250]  eta: 0:28:45  lr: 0.000121  grad: 0.0699 (0.0758)  loss: 0.8362 (0.8326)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [16]  [ 400/6250]  eta: 0:27:50  lr: 0.000121  grad: 0.0657 (0.0744)  loss: 0.8380 (0.8337)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [16]  [ 500/6250]  eta: 0:27:06  lr: 0.000121  grad: 0.0686 (0.0737)  loss: 0.8383 (0.8345)  time: 0.2752  data: 0.0002  max mem: 26157
Train: [16]  [ 600/6250]  eta: 0:26:27  lr: 0.000121  grad: 0.0670 (0.0728)  loss: 0.8307 (0.8348)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [16]  [ 700/6250]  eta: 0:25:51  lr: 0.000121  grad: 0.0736 (0.0727)  loss: 0.8284 (0.8347)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [16]  [ 800/6250]  eta: 0:25:18  lr: 0.000121  grad: 0.0679 (0.0723)  loss: 0.8370 (0.8349)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [16]  [ 900/6250]  eta: 0:24:45  lr: 0.000121  grad: 0.0697 (0.0722)  loss: 0.8359 (0.8347)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [16]  [1000/6250]  eta: 0:24:13  lr: 0.000121  grad: 0.0680 (0.0718)  loss: 0.8297 (0.8347)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [16]  [1100/6250]  eta: 0:23:42  lr: 0.000121  grad: 0.0688 (0.0717)  loss: 0.8326 (0.8344)  time: 0.2685  data: 0.0003  max mem: 26157
Train: [16]  [1200/6250]  eta: 0:23:11  lr: 0.000121  grad: 0.0658 (0.0718)  loss: 0.8308 (0.8341)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [16]  [1300/6250]  eta: 0:22:41  lr: 0.000121  grad: 0.0705 (0.0717)  loss: 0.8300 (0.8340)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [16]  [1400/6250]  eta: 0:22:12  lr: 0.000121  grad: 0.0717 (0.0721)  loss: 0.8290 (0.8336)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [16]  [1500/6250]  eta: 0:21:42  lr: 0.000121  grad: 0.0737 (0.0721)  loss: 0.8307 (0.8334)  time: 0.2681  data: 0.0003  max mem: 26157
Train: [16]  [1600/6250]  eta: 0:21:14  lr: 0.000121  grad: 0.0713 (0.0721)  loss: 0.8304 (0.8331)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [16]  [1700/6250]  eta: 0:20:50  lr: 0.000121  grad: 0.0664 (0.0721)  loss: 0.8331 (0.8329)  time: 0.2941  data: 0.0209  max mem: 26157
Train: [16]  [1800/6250]  eta: 0:20:32  lr: 0.000121  grad: 0.0704 (0.0721)  loss: 0.8286 (0.8327)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [16]  [1900/6250]  eta: 0:20:03  lr: 0.000121  grad: 0.0714 (0.0721)  loss: 0.8291 (0.8325)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [16]  [2000/6250]  eta: 0:19:34  lr: 0.000121  grad: 0.0681 (0.0721)  loss: 0.8341 (0.8324)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [16]  [2100/6250]  eta: 0:19:05  lr: 0.000121  grad: 0.0673 (0.0721)  loss: 0.8270 (0.8323)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [16]  [2200/6250]  eta: 0:18:36  lr: 0.000121  grad: 0.0731 (0.0721)  loss: 0.8288 (0.8322)  time: 0.2693  data: 0.0003  max mem: 26157
Train: [16]  [2300/6250]  eta: 0:18:07  lr: 0.000121  grad: 0.0679 (0.0721)  loss: 0.8302 (0.8321)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [16]  [2400/6250]  eta: 0:17:45  lr: 0.000121  grad: 0.0698 (0.0721)  loss: 0.8306 (0.8321)  time: 0.4632  data: 0.1944  max mem: 26157
Train: [16]  [2500/6250]  eta: 0:17:16  lr: 0.000121  grad: 0.0745 (0.0722)  loss: 0.8263 (0.8319)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [16]  [2600/6250]  eta: 0:16:48  lr: 0.000121  grad: 0.0751 (0.0723)  loss: 0.8297 (0.8318)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [16]  [2700/6250]  eta: 0:16:19  lr: 0.000121  grad: 0.0702 (0.0723)  loss: 0.8309 (0.8317)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [16]  [2800/6250]  eta: 0:15:51  lr: 0.000121  grad: 0.0720 (0.0724)  loss: 0.8199 (0.8315)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [16]  [2900/6250]  eta: 0:15:23  lr: 0.000121  grad: 0.0708 (0.0725)  loss: 0.8258 (0.8313)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [16]  [3000/6250]  eta: 0:14:54  lr: 0.000121  grad: 0.0737 (0.0724)  loss: 0.8265 (0.8311)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [16]  [3100/6250]  eta: 0:14:26  lr: 0.000121  grad: 0.0735 (0.0725)  loss: 0.8242 (0.8310)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [16]  [3200/6250]  eta: 0:13:58  lr: 0.000121  grad: 0.0698 (0.0725)  loss: 0.8269 (0.8309)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [16]  [3300/6250]  eta: 0:13:30  lr: 0.000121  grad: 0.0797 (0.0727)  loss: 0.8182 (0.8308)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [16]  [3400/6250]  eta: 0:13:02  lr: 0.000121  grad: 0.0722 (0.0727)  loss: 0.8215 (0.8306)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [16]  [3500/6250]  eta: 0:12:34  lr: 0.000120  grad: 0.0710 (0.0728)  loss: 0.8277 (0.8305)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [16]  [3600/6250]  eta: 0:12:07  lr: 0.000120  grad: 0.0692 (0.0728)  loss: 0.8325 (0.8305)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [16]  [3700/6250]  eta: 0:11:39  lr: 0.000120  grad: 0.0689 (0.0728)  loss: 0.8262 (0.8304)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [16]  [3800/6250]  eta: 0:11:11  lr: 0.000120  grad: 0.0769 (0.0729)  loss: 0.8264 (0.8303)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [16]  [3900/6250]  eta: 0:10:44  lr: 0.000120  grad: 0.0687 (0.0729)  loss: 0.8277 (0.8303)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [16]  [4000/6250]  eta: 0:10:16  lr: 0.000120  grad: 0.0665 (0.0730)  loss: 0.8287 (0.8302)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [16]  [4100/6250]  eta: 0:09:48  lr: 0.000120  grad: 0.0741 (0.0732)  loss: 0.8214 (0.8301)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [16]  [4200/6250]  eta: 0:09:21  lr: 0.000120  grad: 0.0756 (0.0732)  loss: 0.8249 (0.8300)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [16]  [4300/6250]  eta: 0:08:53  lr: 0.000120  grad: 0.0739 (0.0732)  loss: 0.8251 (0.8300)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [16]  [4400/6250]  eta: 0:08:26  lr: 0.000120  grad: 0.0726 (0.0733)  loss: 0.8247 (0.8299)  time: 0.2745  data: 0.0002  max mem: 26157
Train: [16]  [4500/6250]  eta: 0:07:59  lr: 0.000120  grad: 0.0758 (0.0733)  loss: 0.8286 (0.8298)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [16]  [4600/6250]  eta: 0:07:31  lr: 0.000120  grad: 0.0724 (0.0733)  loss: 0.8284 (0.8298)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [16]  [4700/6250]  eta: 0:07:04  lr: 0.000120  grad: 0.0754 (0.0733)  loss: 0.8310 (0.8298)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [16]  [4800/6250]  eta: 0:06:36  lr: 0.000120  grad: 0.0710 (0.0734)  loss: 0.8285 (0.8297)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [16]  [4900/6250]  eta: 0:06:10  lr: 0.000120  grad: 0.0682 (0.0733)  loss: 0.8263 (0.8297)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [16]  [5000/6250]  eta: 0:05:43  lr: 0.000120  grad: 0.0702 (0.0733)  loss: 0.8266 (0.8297)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [16]  [5100/6250]  eta: 0:05:15  lr: 0.000120  grad: 0.0715 (0.0733)  loss: 0.8314 (0.8297)  time: 0.2696  data: 0.0003  max mem: 26157
Train: [16]  [5200/6250]  eta: 0:04:47  lr: 0.000120  grad: 0.0682 (0.0733)  loss: 0.8265 (0.8297)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [16]  [5300/6250]  eta: 0:04:20  lr: 0.000120  grad: 0.0722 (0.0734)  loss: 0.8316 (0.8297)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [16]  [5400/6250]  eta: 0:03:52  lr: 0.000120  grad: 0.0708 (0.0733)  loss: 0.8301 (0.8297)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [16]  [5500/6250]  eta: 0:03:25  lr: 0.000120  grad: 0.0693 (0.0733)  loss: 0.8308 (0.8298)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [16]  [5600/6250]  eta: 0:02:57  lr: 0.000120  grad: 0.0721 (0.0733)  loss: 0.8324 (0.8298)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [16]  [5700/6250]  eta: 0:02:30  lr: 0.000120  grad: 0.0758 (0.0733)  loss: 0.8274 (0.8298)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [16]  [5800/6250]  eta: 0:02:03  lr: 0.000120  grad: 0.0760 (0.0733)  loss: 0.8248 (0.8298)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [16]  [5900/6250]  eta: 0:01:35  lr: 0.000120  grad: 0.0712 (0.0733)  loss: 0.8291 (0.8298)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [16]  [6000/6250]  eta: 0:01:08  lr: 0.000120  grad: 0.0763 (0.0734)  loss: 0.8234 (0.8297)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [16]  [6100/6250]  eta: 0:00:41  lr: 0.000120  grad: 0.0763 (0.0734)  loss: 0.8275 (0.8297)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [16]  [6200/6250]  eta: 0:00:13  lr: 0.000120  grad: 0.0684 (0.0734)  loss: 0.8336 (0.8297)  time: 0.3009  data: 0.0299  max mem: 26157
Train: [16]  [6249/6250]  eta: 0:00:00  lr: 0.000120  grad: 0.0773 (0.0735)  loss: 0.8275 (0.8297)  time: 0.2785  data: 0.0002  max mem: 26157
Train: [16] Total time: 0:28:36 (0.2746 s / it)
Averaged stats: lr: 0.000120  grad: 0.0773 (0.0735)  loss: 0.8275 (0.8297)
Eval (hcp-train-subset): [16]  [ 0/62]  eta: 0:05:25  loss: 0.8519 (0.8519)  time: 5.2443  data: 5.1603  max mem: 26157
Eval (hcp-train-subset): [16]  [61/62]  eta: 0:00:00  loss: 0.8376 (0.8399)  time: 0.1137  data: 0.0288  max mem: 26157
Eval (hcp-train-subset): [16] Total time: 0:00:12 (0.2084 s / it)
Averaged stats (hcp-train-subset): loss: 0.8376 (0.8399)
Making plots (hcp-train-subset): example=57
Eval (hcp-val): [16]  [ 0/62]  eta: 0:05:03  loss: 0.8335 (0.8335)  time: 4.8919  data: 4.8075  max mem: 26157
Eval (hcp-val): [16]  [61/62]  eta: 0:00:00  loss: 0.8340 (0.8358)  time: 0.1336  data: 0.0506  max mem: 26157
Eval (hcp-val): [16] Total time: 0:00:12 (0.2081 s / it)
Averaged stats (hcp-val): loss: 0.8340 (0.8358)
Making plots (hcp-val): example=27
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [17]  [   0/6250]  eta: 10:50:37  lr: 0.000120  grad: 0.0792 (0.0792)  loss: 0.8202 (0.8202)  time: 6.2460  data: 5.9654  max mem: 26157
Train: [17]  [ 100/6250]  eta: 0:34:01  lr: 0.000120  grad: 0.0794 (0.0849)  loss: 0.8271 (0.8303)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [17]  [ 200/6250]  eta: 0:30:27  lr: 0.000120  grad: 0.0769 (0.0827)  loss: 0.8208 (0.8254)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [17]  [ 300/6250]  eta: 0:28:58  lr: 0.000120  grad: 0.0723 (0.0802)  loss: 0.8317 (0.8254)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [17]  [ 400/6250]  eta: 0:28:00  lr: 0.000120  grad: 0.0707 (0.0789)  loss: 0.8291 (0.8259)  time: 0.2713  data: 0.0003  max mem: 26157
Train: [17]  [ 500/6250]  eta: 0:27:15  lr: 0.000120  grad: 0.0712 (0.0787)  loss: 0.8292 (0.8264)  time: 0.2721  data: 0.0003  max mem: 26157
Train: [17]  [ 600/6250]  eta: 0:26:34  lr: 0.000120  grad: 0.0691 (0.0777)  loss: 0.8286 (0.8269)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [17]  [ 700/6250]  eta: 0:25:56  lr: 0.000120  grad: 0.0651 (0.0770)  loss: 0.8336 (0.8274)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [17]  [ 800/6250]  eta: 0:25:23  lr: 0.000120  grad: 0.0684 (0.0763)  loss: 0.8271 (0.8278)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [17]  [ 900/6250]  eta: 0:24:50  lr: 0.000120  grad: 0.0688 (0.0759)  loss: 0.8291 (0.8279)  time: 0.2744  data: 0.0002  max mem: 26157
Train: [17]  [1000/6250]  eta: 0:24:19  lr: 0.000120  grad: 0.0713 (0.0755)  loss: 0.8315 (0.8279)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [17]  [1100/6250]  eta: 0:23:48  lr: 0.000120  grad: 0.0648 (0.0752)  loss: 0.8320 (0.8279)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [17]  [1200/6250]  eta: 0:23:17  lr: 0.000120  grad: 0.0664 (0.0750)  loss: 0.8307 (0.8278)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [17]  [1300/6250]  eta: 0:22:47  lr: 0.000120  grad: 0.0668 (0.0747)  loss: 0.8336 (0.8279)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [17]  [1400/6250]  eta: 0:22:17  lr: 0.000120  grad: 0.0701 (0.0745)  loss: 0.8286 (0.8279)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [17]  [1500/6250]  eta: 0:22:03  lr: 0.000120  grad: 0.0722 (0.0743)  loss: 0.8255 (0.8279)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [17]  [1600/6250]  eta: 0:21:33  lr: 0.000120  grad: 0.0685 (0.0742)  loss: 0.8213 (0.8279)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [17]  [1700/6250]  eta: 0:21:02  lr: 0.000120  grad: 0.0744 (0.0741)  loss: 0.8286 (0.8280)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [17]  [1800/6250]  eta: 0:20:33  lr: 0.000120  grad: 0.0759 (0.0740)  loss: 0.8227 (0.8279)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [17]  [1900/6250]  eta: 0:20:03  lr: 0.000120  grad: 0.0661 (0.0739)  loss: 0.8342 (0.8280)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [17]  [2000/6250]  eta: 0:19:34  lr: 0.000120  grad: 0.0725 (0.0738)  loss: 0.8272 (0.8281)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [17]  [2100/6250]  eta: 0:19:06  lr: 0.000120  grad: 0.0733 (0.0738)  loss: 0.8307 (0.8282)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [17]  [2200/6250]  eta: 0:18:37  lr: 0.000120  grad: 0.0716 (0.0737)  loss: 0.8280 (0.8282)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [17]  [2300/6250]  eta: 0:18:08  lr: 0.000120  grad: 0.0656 (0.0735)  loss: 0.8364 (0.8284)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [17]  [2400/6250]  eta: 0:17:40  lr: 0.000120  grad: 0.0694 (0.0734)  loss: 0.8321 (0.8286)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [17]  [2500/6250]  eta: 0:17:11  lr: 0.000120  grad: 0.0697 (0.0733)  loss: 0.8335 (0.8287)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [17]  [2600/6250]  eta: 0:16:43  lr: 0.000120  grad: 0.0750 (0.0732)  loss: 0.8286 (0.8288)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [17]  [2700/6250]  eta: 0:16:15  lr: 0.000120  grad: 0.0653 (0.0732)  loss: 0.8316 (0.8289)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [17]  [2800/6250]  eta: 0:15:46  lr: 0.000120  grad: 0.0670 (0.0730)  loss: 0.8321 (0.8290)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [17]  [2900/6250]  eta: 0:15:18  lr: 0.000120  grad: 0.0683 (0.0728)  loss: 0.8287 (0.8291)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [17]  [3000/6250]  eta: 0:14:50  lr: 0.000120  grad: 0.0694 (0.0728)  loss: 0.8249 (0.8291)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [17]  [3100/6250]  eta: 0:14:23  lr: 0.000120  grad: 0.0703 (0.0727)  loss: 0.8350 (0.8292)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [17]  [3200/6250]  eta: 0:14:02  lr: 0.000120  grad: 0.0669 (0.0727)  loss: 0.8346 (0.8293)  time: 0.2745  data: 0.0002  max mem: 26157
Train: [17]  [3300/6250]  eta: 0:13:34  lr: 0.000120  grad: 0.0717 (0.0726)  loss: 0.8320 (0.8293)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [17]  [3400/6250]  eta: 0:13:12  lr: 0.000120  grad: 0.0697 (0.0727)  loss: 0.8290 (0.8293)  time: 0.2707  data: 0.0003  max mem: 26157
Train: [17]  [3500/6250]  eta: 0:12:43  lr: 0.000120  grad: 0.0676 (0.0726)  loss: 0.8358 (0.8294)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [17]  [3600/6250]  eta: 0:12:15  lr: 0.000120  grad: 0.0723 (0.0727)  loss: 0.8321 (0.8295)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [17]  [3700/6250]  eta: 0:11:47  lr: 0.000120  grad: 0.0731 (0.0727)  loss: 0.8291 (0.8295)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [17]  [3800/6250]  eta: 0:11:18  lr: 0.000120  grad: 0.0682 (0.0727)  loss: 0.8304 (0.8295)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [17]  [3900/6250]  eta: 0:10:50  lr: 0.000120  grad: 0.0671 (0.0726)  loss: 0.8288 (0.8295)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [17]  [4000/6250]  eta: 0:10:22  lr: 0.000120  grad: 0.0690 (0.0726)  loss: 0.8297 (0.8295)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [17]  [4100/6250]  eta: 0:09:54  lr: 0.000120  grad: 0.0719 (0.0726)  loss: 0.8241 (0.8295)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [17]  [4200/6250]  eta: 0:09:26  lr: 0.000120  grad: 0.0714 (0.0727)  loss: 0.8263 (0.8295)  time: 0.2699  data: 0.0003  max mem: 26157
Train: [17]  [4300/6250]  eta: 0:08:58  lr: 0.000120  grad: 0.0656 (0.0726)  loss: 0.8385 (0.8296)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [17]  [4400/6250]  eta: 0:08:31  lr: 0.000120  grad: 0.0705 (0.0727)  loss: 0.8306 (0.8296)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [17]  [4500/6250]  eta: 0:08:03  lr: 0.000120  grad: 0.0733 (0.0727)  loss: 0.8277 (0.8296)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [17]  [4600/6250]  eta: 0:07:35  lr: 0.000120  grad: 0.0735 (0.0727)  loss: 0.8292 (0.8297)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [17]  [4700/6250]  eta: 0:07:07  lr: 0.000120  grad: 0.0698 (0.0727)  loss: 0.8313 (0.8296)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [17]  [4800/6250]  eta: 0:06:39  lr: 0.000120  grad: 0.0706 (0.0728)  loss: 0.8297 (0.8297)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [17]  [4900/6250]  eta: 0:06:11  lr: 0.000119  grad: 0.0739 (0.0729)  loss: 0.8319 (0.8297)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [17]  [5000/6250]  eta: 0:05:44  lr: 0.000119  grad: 0.0742 (0.0729)  loss: 0.8227 (0.8297)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [17]  [5100/6250]  eta: 0:05:16  lr: 0.000119  grad: 0.0756 (0.0730)  loss: 0.8262 (0.8297)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [17]  [5200/6250]  eta: 0:04:49  lr: 0.000119  grad: 0.0723 (0.0731)  loss: 0.8240 (0.8296)  time: 0.3048  data: 0.0348  max mem: 26157
Train: [17]  [5300/6250]  eta: 0:04:22  lr: 0.000119  grad: 0.0769 (0.0732)  loss: 0.8285 (0.8296)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [17]  [5400/6250]  eta: 0:03:54  lr: 0.000119  grad: 0.0797 (0.0733)  loss: 0.8227 (0.8296)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [17]  [5500/6250]  eta: 0:03:26  lr: 0.000119  grad: 0.0755 (0.0733)  loss: 0.8267 (0.8295)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [17]  [5600/6250]  eta: 0:02:59  lr: 0.000119  grad: 0.0763 (0.0733)  loss: 0.8296 (0.8295)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [17]  [5700/6250]  eta: 0:02:31  lr: 0.000119  grad: 0.0759 (0.0734)  loss: 0.8253 (0.8294)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [17]  [5800/6250]  eta: 0:02:04  lr: 0.000119  grad: 0.0754 (0.0734)  loss: 0.8256 (0.8293)  time: 0.2726  data: 0.0003  max mem: 26157
Train: [17]  [5900/6250]  eta: 0:01:36  lr: 0.000119  grad: 0.0762 (0.0735)  loss: 0.8249 (0.8293)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [17]  [6000/6250]  eta: 0:01:08  lr: 0.000119  grad: 0.0754 (0.0735)  loss: 0.8263 (0.8292)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [17]  [6100/6250]  eta: 0:00:41  lr: 0.000119  grad: 0.0739 (0.0736)  loss: 0.8205 (0.8291)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [17]  [6200/6250]  eta: 0:00:13  lr: 0.000119  grad: 0.0726 (0.0737)  loss: 0.8273 (0.8290)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [17]  [6249/6250]  eta: 0:00:00  lr: 0.000119  grad: 0.0726 (0.0737)  loss: 0.8253 (0.8290)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [17] Total time: 0:28:55 (0.2777 s / it)
Averaged stats: lr: 0.000119  grad: 0.0726 (0.0737)  loss: 0.8253 (0.8290)
Eval (hcp-train-subset): [17]  [ 0/62]  eta: 0:05:19  loss: 0.8488 (0.8488)  time: 5.1472  data: 5.0617  max mem: 26157
Eval (hcp-train-subset): [17]  [61/62]  eta: 0:00:00  loss: 0.8359 (0.8387)  time: 0.1338  data: 0.0510  max mem: 26157
Eval (hcp-train-subset): [17] Total time: 0:00:13 (0.2104 s / it)
Averaged stats (hcp-train-subset): loss: 0.8359 (0.8387)
Making plots (hcp-train-subset): example=3
Eval (hcp-val): [17]  [ 0/62]  eta: 0:03:20  loss: 0.8339 (0.8339)  time: 3.2295  data: 3.1248  max mem: 26157
Eval (hcp-val): [17]  [61/62]  eta: 0:00:00  loss: 0.8329 (0.8352)  time: 0.0829  data: 0.0002  max mem: 26157
Eval (hcp-val): [17] Total time: 0:00:17 (0.2872 s / it)
Averaged stats (hcp-val): loss: 0.8329 (0.8352)
Making plots (hcp-val): example=38
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [18]  [   0/6250]  eta: 9:22:26  lr: 0.000119  grad: 0.0795 (0.0795)  loss: 0.8629 (0.8629)  time: 5.3995  data: 5.1095  max mem: 26157
Train: [18]  [ 100/6250]  eta: 0:34:11  lr: 0.000119  grad: 0.0775 (0.0825)  loss: 0.8279 (0.8325)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [18]  [ 200/6250]  eta: 0:30:29  lr: 0.000119  grad: 0.0745 (0.0792)  loss: 0.8252 (0.8302)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [18]  [ 300/6250]  eta: 0:28:55  lr: 0.000119  grad: 0.0714 (0.0772)  loss: 0.8288 (0.8300)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [18]  [ 400/6250]  eta: 0:27:56  lr: 0.000119  grad: 0.0706 (0.0763)  loss: 0.8320 (0.8304)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [18]  [ 500/6250]  eta: 0:27:10  lr: 0.000119  grad: 0.0710 (0.0755)  loss: 0.8309 (0.8305)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [18]  [ 600/6250]  eta: 0:26:30  lr: 0.000119  grad: 0.0677 (0.0747)  loss: 0.8277 (0.8304)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [18]  [ 700/6250]  eta: 0:25:53  lr: 0.000119  grad: 0.0697 (0.0744)  loss: 0.8281 (0.8302)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [18]  [ 800/6250]  eta: 0:25:18  lr: 0.000119  grad: 0.0667 (0.0741)  loss: 0.8256 (0.8298)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [18]  [ 900/6250]  eta: 0:24:45  lr: 0.000119  grad: 0.0691 (0.0737)  loss: 0.8272 (0.8298)  time: 0.2712  data: 0.0003  max mem: 26157
Train: [18]  [1000/6250]  eta: 0:24:14  lr: 0.000119  grad: 0.0719 (0.0735)  loss: 0.8304 (0.8297)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [18]  [1100/6250]  eta: 0:23:43  lr: 0.000119  grad: 0.0680 (0.0734)  loss: 0.8339 (0.8295)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [18]  [1200/6250]  eta: 0:23:16  lr: 0.000119  grad: 0.0727 (0.0732)  loss: 0.8262 (0.8294)  time: 0.2709  data: 0.0003  max mem: 26157
Train: [18]  [1300/6250]  eta: 0:22:54  lr: 0.000119  grad: 0.0718 (0.0732)  loss: 0.8251 (0.8292)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [18]  [1400/6250]  eta: 0:22:24  lr: 0.000119  grad: 0.0727 (0.0731)  loss: 0.8232 (0.8290)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [18]  [1500/6250]  eta: 0:21:54  lr: 0.000119  grad: 0.0685 (0.0731)  loss: 0.8300 (0.8287)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [18]  [1600/6250]  eta: 0:21:25  lr: 0.000119  grad: 0.0683 (0.0730)  loss: 0.8255 (0.8285)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [18]  [1700/6250]  eta: 0:20:56  lr: 0.000119  grad: 0.0735 (0.0729)  loss: 0.8262 (0.8284)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [18]  [1800/6250]  eta: 0:20:27  lr: 0.000119  grad: 0.0728 (0.0728)  loss: 0.8213 (0.8283)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [18]  [1900/6250]  eta: 0:19:58  lr: 0.000119  grad: 0.0775 (0.0730)  loss: 0.8194 (0.8280)  time: 0.2748  data: 0.0002  max mem: 26157
Train: [18]  [2000/6250]  eta: 0:19:39  lr: 0.000119  grad: 0.0756 (0.0731)  loss: 0.8223 (0.8278)  time: 0.3829  data: 0.1049  max mem: 26157
Train: [18]  [2100/6250]  eta: 0:19:10  lr: 0.000119  grad: 0.0748 (0.0732)  loss: 0.8303 (0.8277)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [18]  [2200/6250]  eta: 0:18:45  lr: 0.000119  grad: 0.0750 (0.0732)  loss: 0.8269 (0.8276)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [18]  [2300/6250]  eta: 0:18:16  lr: 0.000119  grad: 0.0747 (0.0733)  loss: 0.8300 (0.8275)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [18]  [2400/6250]  eta: 0:17:47  lr: 0.000119  grad: 0.0739 (0.0733)  loss: 0.8245 (0.8275)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [18]  [2500/6250]  eta: 0:17:18  lr: 0.000119  grad: 0.0699 (0.0732)  loss: 0.8268 (0.8276)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [18]  [2600/6250]  eta: 0:16:50  lr: 0.000119  grad: 0.0660 (0.0731)  loss: 0.8326 (0.8276)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [18]  [2700/6250]  eta: 0:16:21  lr: 0.000119  grad: 0.0700 (0.0731)  loss: 0.8270 (0.8277)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [18]  [2800/6250]  eta: 0:15:53  lr: 0.000119  grad: 0.0674 (0.0731)  loss: 0.8275 (0.8277)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [18]  [2900/6250]  eta: 0:15:25  lr: 0.000119  grad: 0.0675 (0.0730)  loss: 0.8329 (0.8278)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [18]  [3000/6250]  eta: 0:14:56  lr: 0.000119  grad: 0.0737 (0.0730)  loss: 0.8249 (0.8278)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [18]  [3100/6250]  eta: 0:14:28  lr: 0.000119  grad: 0.0675 (0.0729)  loss: 0.8281 (0.8279)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [18]  [3200/6250]  eta: 0:14:00  lr: 0.000119  grad: 0.0702 (0.0728)  loss: 0.8252 (0.8279)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [18]  [3300/6250]  eta: 0:13:32  lr: 0.000119  grad: 0.0702 (0.0727)  loss: 0.8248 (0.8279)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [18]  [3400/6250]  eta: 0:13:04  lr: 0.000119  grad: 0.0677 (0.0727)  loss: 0.8272 (0.8279)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [18]  [3500/6250]  eta: 0:12:36  lr: 0.000119  grad: 0.0743 (0.0728)  loss: 0.8268 (0.8278)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [18]  [3600/6250]  eta: 0:12:09  lr: 0.000119  grad: 0.0715 (0.0729)  loss: 0.8276 (0.8278)  time: 0.2705  data: 0.0001  max mem: 26157
Train: [18]  [3700/6250]  eta: 0:11:41  lr: 0.000119  grad: 0.0684 (0.0729)  loss: 0.8271 (0.8278)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [18]  [3800/6250]  eta: 0:11:14  lr: 0.000119  grad: 0.0787 (0.0731)  loss: 0.8266 (0.8276)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [18]  [3900/6250]  eta: 0:10:46  lr: 0.000119  grad: 0.0764 (0.0732)  loss: 0.8289 (0.8276)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [18]  [4000/6250]  eta: 0:10:18  lr: 0.000119  grad: 0.0770 (0.0732)  loss: 0.8208 (0.8276)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [18]  [4100/6250]  eta: 0:09:50  lr: 0.000119  grad: 0.0733 (0.0733)  loss: 0.8258 (0.8276)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [18]  [4200/6250]  eta: 0:09:22  lr: 0.000119  grad: 0.0739 (0.0733)  loss: 0.8237 (0.8276)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [18]  [4300/6250]  eta: 0:08:55  lr: 0.000119  grad: 0.0745 (0.0734)  loss: 0.8222 (0.8276)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [18]  [4400/6250]  eta: 0:08:27  lr: 0.000119  grad: 0.0724 (0.0734)  loss: 0.8243 (0.8276)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [18]  [4500/6250]  eta: 0:07:59  lr: 0.000119  grad: 0.0756 (0.0734)  loss: 0.8283 (0.8275)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [18]  [4600/6250]  eta: 0:07:33  lr: 0.000119  grad: 0.0711 (0.0735)  loss: 0.8289 (0.8275)  time: 0.2765  data: 0.0002  max mem: 26157
Train: [18]  [4700/6250]  eta: 0:07:05  lr: 0.000119  grad: 0.0745 (0.0735)  loss: 0.8239 (0.8275)  time: 0.2762  data: 0.0002  max mem: 26157
Train: [18]  [4800/6250]  eta: 0:06:38  lr: 0.000119  grad: 0.0700 (0.0735)  loss: 0.8270 (0.8276)  time: 0.2693  data: 0.0001  max mem: 26157
Train: [18]  [4900/6250]  eta: 0:06:10  lr: 0.000119  grad: 0.0716 (0.0736)  loss: 0.8269 (0.8276)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [18]  [5000/6250]  eta: 0:05:43  lr: 0.000119  grad: 0.0686 (0.0736)  loss: 0.8314 (0.8276)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [18]  [5100/6250]  eta: 0:05:15  lr: 0.000119  grad: 0.0702 (0.0735)  loss: 0.8284 (0.8276)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [18]  [5200/6250]  eta: 0:04:47  lr: 0.000119  grad: 0.0782 (0.0736)  loss: 0.8311 (0.8277)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [18]  [5300/6250]  eta: 0:04:20  lr: 0.000119  grad: 0.0740 (0.0736)  loss: 0.8274 (0.8277)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [18]  [5400/6250]  eta: 0:03:52  lr: 0.000119  grad: 0.0721 (0.0736)  loss: 0.8299 (0.8277)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [18]  [5500/6250]  eta: 0:03:25  lr: 0.000119  grad: 0.0767 (0.0737)  loss: 0.8243 (0.8276)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [18]  [5600/6250]  eta: 0:02:58  lr: 0.000119  grad: 0.0714 (0.0737)  loss: 0.8281 (0.8276)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [18]  [5700/6250]  eta: 0:02:30  lr: 0.000119  grad: 0.0708 (0.0738)  loss: 0.8236 (0.8276)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [18]  [5800/6250]  eta: 0:02:03  lr: 0.000118  grad: 0.0735 (0.0738)  loss: 0.8277 (0.8276)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [18]  [5900/6250]  eta: 0:01:35  lr: 0.000118  grad: 0.0747 (0.0738)  loss: 0.8276 (0.8275)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [18]  [6000/6250]  eta: 0:01:08  lr: 0.000118  grad: 0.0775 (0.0739)  loss: 0.8194 (0.8275)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [18]  [6100/6250]  eta: 0:00:41  lr: 0.000118  grad: 0.0746 (0.0739)  loss: 0.8250 (0.8274)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [18]  [6200/6250]  eta: 0:00:13  lr: 0.000118  grad: 0.0739 (0.0739)  loss: 0.8265 (0.8274)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [18]  [6249/6250]  eta: 0:00:00  lr: 0.000118  grad: 0.0728 (0.0739)  loss: 0.8289 (0.8274)  time: 0.2694  data: 0.0001  max mem: 26157
Train: [18] Total time: 0:28:36 (0.2746 s / it)
Averaged stats: lr: 0.000118  grad: 0.0728 (0.0739)  loss: 0.8289 (0.8274)
Eval (hcp-train-subset): [18]  [ 0/62]  eta: 0:05:58  loss: 0.8508 (0.8508)  time: 5.7747  data: 5.6897  max mem: 26157
Eval (hcp-train-subset): [18]  [61/62]  eta: 0:00:00  loss: 0.8359 (0.8387)  time: 0.1437  data: 0.0610  max mem: 26157
Eval (hcp-train-subset): [18] Total time: 0:00:14 (0.2413 s / it)
Averaged stats (hcp-train-subset): loss: 0.8359 (0.8387)
Making plots (hcp-train-subset): example=2
Eval (hcp-val): [18]  [ 0/62]  eta: 0:04:56  loss: 0.8324 (0.8324)  time: 4.7857  data: 4.6945  max mem: 26157
Eval (hcp-val): [18]  [61/62]  eta: 0:00:00  loss: 0.8323 (0.8342)  time: 0.1583  data: 0.0751  max mem: 26157
Eval (hcp-val): [18] Total time: 0:00:14 (0.2357 s / it)
Averaged stats (hcp-val): loss: 0.8323 (0.8342)
Making plots (hcp-val): example=25
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [19]  [   0/6250]  eta: 12:09:00  lr: 0.000118  grad: 0.1213 (0.1213)  loss: 0.8194 (0.8194)  time: 6.9985  data: 6.7215  max mem: 26157
Train: [19]  [ 100/6250]  eta: 0:34:42  lr: 0.000118  grad: 0.0772 (0.0833)  loss: 0.8387 (0.8401)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [19]  [ 200/6250]  eta: 0:30:42  lr: 0.000118  grad: 0.0718 (0.0796)  loss: 0.8327 (0.8369)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [19]  [ 300/6250]  eta: 0:29:03  lr: 0.000118  grad: 0.0658 (0.0783)  loss: 0.8358 (0.8350)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [19]  [ 400/6250]  eta: 0:28:19  lr: 0.000118  grad: 0.0709 (0.0786)  loss: 0.8301 (0.8336)  time: 0.3333  data: 0.0002  max mem: 26157
Train: [19]  [ 500/6250]  eta: 0:27:28  lr: 0.000118  grad: 0.0713 (0.0778)  loss: 0.8265 (0.8325)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [19]  [ 600/6250]  eta: 0:26:44  lr: 0.000118  grad: 0.0708 (0.0768)  loss: 0.8275 (0.8319)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [19]  [ 700/6250]  eta: 0:26:05  lr: 0.000118  grad: 0.0720 (0.0763)  loss: 0.8312 (0.8317)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [19]  [ 800/6250]  eta: 0:25:29  lr: 0.000118  grad: 0.0736 (0.0759)  loss: 0.8360 (0.8317)  time: 0.2768  data: 0.0002  max mem: 26157
Train: [19]  [ 900/6250]  eta: 0:24:59  lr: 0.000118  grad: 0.0648 (0.0753)  loss: 0.8321 (0.8318)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [19]  [1000/6250]  eta: 0:24:27  lr: 0.000118  grad: 0.0651 (0.0748)  loss: 0.8288 (0.8318)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [19]  [1100/6250]  eta: 0:23:56  lr: 0.000118  grad: 0.0703 (0.0746)  loss: 0.8286 (0.8317)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [19]  [1200/6250]  eta: 0:23:25  lr: 0.000118  grad: 0.0746 (0.0745)  loss: 0.8264 (0.8314)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [19]  [1300/6250]  eta: 0:22:53  lr: 0.000118  grad: 0.0735 (0.0744)  loss: 0.8243 (0.8311)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [19]  [1400/6250]  eta: 0:22:23  lr: 0.000118  grad: 0.0722 (0.0744)  loss: 0.8246 (0.8308)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [19]  [1500/6250]  eta: 0:21:52  lr: 0.000118  grad: 0.0761 (0.0745)  loss: 0.8229 (0.8305)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [19]  [1600/6250]  eta: 0:21:22  lr: 0.000118  grad: 0.0712 (0.0745)  loss: 0.8311 (0.8303)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [19]  [1700/6250]  eta: 0:20:53  lr: 0.000118  grad: 0.0687 (0.0745)  loss: 0.8267 (0.8303)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [19]  [1800/6250]  eta: 0:20:24  lr: 0.000118  grad: 0.0741 (0.0745)  loss: 0.8281 (0.8301)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [19]  [1900/6250]  eta: 0:19:55  lr: 0.000118  grad: 0.0782 (0.0747)  loss: 0.8217 (0.8298)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [19]  [2000/6250]  eta: 0:19:27  lr: 0.000118  grad: 0.0737 (0.0749)  loss: 0.8305 (0.8297)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [19]  [2100/6250]  eta: 0:18:58  lr: 0.000118  grad: 0.0720 (0.0749)  loss: 0.8267 (0.8296)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [19]  [2200/6250]  eta: 0:18:30  lr: 0.000118  grad: 0.0757 (0.0750)  loss: 0.8232 (0.8295)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [19]  [2300/6250]  eta: 0:18:02  lr: 0.000118  grad: 0.0735 (0.0750)  loss: 0.8248 (0.8293)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [19]  [2400/6250]  eta: 0:17:34  lr: 0.000118  grad: 0.0748 (0.0751)  loss: 0.8240 (0.8291)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [19]  [2500/6250]  eta: 0:17:06  lr: 0.000118  grad: 0.0731 (0.0753)  loss: 0.8177 (0.8288)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [19]  [2600/6250]  eta: 0:16:38  lr: 0.000118  grad: 0.0752 (0.0754)  loss: 0.8189 (0.8286)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [19]  [2700/6250]  eta: 0:16:10  lr: 0.000118  grad: 0.0718 (0.0754)  loss: 0.8255 (0.8284)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [19]  [2800/6250]  eta: 0:15:42  lr: 0.000118  grad: 0.0746 (0.0755)  loss: 0.8204 (0.8283)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [19]  [2900/6250]  eta: 0:15:15  lr: 0.000118  grad: 0.0773 (0.0755)  loss: 0.8233 (0.8281)  time: 0.2702  data: 0.0003  max mem: 26157
Train: [19]  [3000/6250]  eta: 0:14:47  lr: 0.000118  grad: 0.0739 (0.0755)  loss: 0.8280 (0.8279)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [19]  [3100/6250]  eta: 0:14:19  lr: 0.000118  grad: 0.0736 (0.0756)  loss: 0.8311 (0.8279)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [19]  [3200/6250]  eta: 0:13:52  lr: 0.000118  grad: 0.0756 (0.0756)  loss: 0.8252 (0.8277)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [19]  [3300/6250]  eta: 0:13:26  lr: 0.000118  grad: 0.0754 (0.0756)  loss: 0.8218 (0.8277)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [19]  [3400/6250]  eta: 0:12:58  lr: 0.000118  grad: 0.0701 (0.0755)  loss: 0.8263 (0.8276)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [19]  [3500/6250]  eta: 0:12:31  lr: 0.000118  grad: 0.0736 (0.0755)  loss: 0.8268 (0.8275)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [19]  [3600/6250]  eta: 0:12:03  lr: 0.000118  grad: 0.0693 (0.0755)  loss: 0.8287 (0.8275)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [19]  [3700/6250]  eta: 0:11:35  lr: 0.000118  grad: 0.0736 (0.0754)  loss: 0.8180 (0.8274)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [19]  [3800/6250]  eta: 0:11:08  lr: 0.000118  grad: 0.0745 (0.0755)  loss: 0.8292 (0.8274)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [19]  [3900/6250]  eta: 0:10:41  lr: 0.000118  grad: 0.0716 (0.0754)  loss: 0.8244 (0.8273)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [19]  [4000/6250]  eta: 0:10:13  lr: 0.000118  grad: 0.0757 (0.0756)  loss: 0.8267 (0.8272)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [19]  [4100/6250]  eta: 0:09:46  lr: 0.000118  grad: 0.0797 (0.0757)  loss: 0.8253 (0.8271)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [19]  [4200/6250]  eta: 0:09:18  lr: 0.000118  grad: 0.0768 (0.0759)  loss: 0.8252 (0.8271)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [19]  [4300/6250]  eta: 0:08:51  lr: 0.000118  grad: 0.0709 (0.0759)  loss: 0.8280 (0.8270)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [19]  [4400/6250]  eta: 0:08:23  lr: 0.000118  grad: 0.0756 (0.0760)  loss: 0.8199 (0.8270)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [19]  [4500/6250]  eta: 0:07:56  lr: 0.000118  grad: 0.0708 (0.0760)  loss: 0.8219 (0.8269)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [19]  [4600/6250]  eta: 0:07:29  lr: 0.000118  grad: 0.0771 (0.0761)  loss: 0.8249 (0.8269)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [19]  [4700/6250]  eta: 0:07:02  lr: 0.000118  grad: 0.0748 (0.0761)  loss: 0.8235 (0.8268)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [19]  [4800/6250]  eta: 0:06:34  lr: 0.000118  grad: 0.0789 (0.0761)  loss: 0.8197 (0.8267)  time: 0.2760  data: 0.0002  max mem: 26157
Train: [19]  [4900/6250]  eta: 0:06:09  lr: 0.000118  grad: 0.0739 (0.0762)  loss: 0.8180 (0.8267)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [19]  [5000/6250]  eta: 0:05:41  lr: 0.000118  grad: 0.0747 (0.0763)  loss: 0.8209 (0.8266)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [19]  [5100/6250]  eta: 0:05:14  lr: 0.000118  grad: 0.0718 (0.0763)  loss: 0.8223 (0.8266)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [19]  [5200/6250]  eta: 0:04:47  lr: 0.000118  grad: 0.0763 (0.0762)  loss: 0.8233 (0.8265)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [19]  [5300/6250]  eta: 0:04:19  lr: 0.000118  grad: 0.0739 (0.0762)  loss: 0.8227 (0.8265)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [19]  [5400/6250]  eta: 0:03:52  lr: 0.000118  grad: 0.0716 (0.0762)  loss: 0.8223 (0.8265)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [19]  [5500/6250]  eta: 0:03:25  lr: 0.000118  grad: 0.0752 (0.0762)  loss: 0.8256 (0.8264)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [19]  [5600/6250]  eta: 0:02:57  lr: 0.000118  grad: 0.0783 (0.0762)  loss: 0.8230 (0.8264)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [19]  [5700/6250]  eta: 0:02:30  lr: 0.000118  grad: 0.0754 (0.0762)  loss: 0.8231 (0.8264)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [19]  [5800/6250]  eta: 0:02:02  lr: 0.000118  grad: 0.0736 (0.0762)  loss: 0.8261 (0.8264)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [19]  [5900/6250]  eta: 0:01:35  lr: 0.000118  grad: 0.0727 (0.0762)  loss: 0.8267 (0.8264)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [19]  [6000/6250]  eta: 0:01:08  lr: 0.000118  grad: 0.0760 (0.0762)  loss: 0.8227 (0.8264)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [19]  [6100/6250]  eta: 0:00:41  lr: 0.000117  grad: 0.0723 (0.0761)  loss: 0.8297 (0.8264)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [19]  [6200/6250]  eta: 0:00:13  lr: 0.000117  grad: 0.0786 (0.0762)  loss: 0.8294 (0.8263)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [19]  [6249/6250]  eta: 0:00:00  lr: 0.000117  grad: 0.0793 (0.0762)  loss: 0.8290 (0.8263)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [19] Total time: 0:28:38 (0.2749 s / it)
Averaged stats: lr: 0.000117  grad: 0.0793 (0.0762)  loss: 0.8290 (0.8263)
Eval (hcp-train-subset): [19]  [ 0/62]  eta: 0:04:24  loss: 0.8451 (0.8451)  time: 4.2592  data: 4.1388  max mem: 26157
Eval (hcp-train-subset): [19]  [61/62]  eta: 0:00:00  loss: 0.8352 (0.8380)  time: 0.1435  data: 0.0607  max mem: 26157
Eval (hcp-train-subset): [19] Total time: 0:00:13 (0.2134 s / it)
Averaged stats (hcp-train-subset): loss: 0.8352 (0.8380)
Making plots (hcp-train-subset): example=18
Eval (hcp-val): [19]  [ 0/62]  eta: 0:05:35  loss: 0.8316 (0.8316)  time: 5.4181  data: 5.3340  max mem: 26157
Eval (hcp-val): [19]  [61/62]  eta: 0:00:00  loss: 0.8323 (0.8337)  time: 0.1216  data: 0.0384  max mem: 26157
Eval (hcp-val): [19] Total time: 0:00:12 (0.2043 s / it)
Averaged stats (hcp-val): loss: 0.8323 (0.8337)
Making plots (hcp-val): example=18
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [20]  [   0/6250]  eta: 6:28:51  lr: 0.000117  grad: 0.0507 (0.0507)  loss: 0.8646 (0.8646)  time: 3.7330  data: 3.3768  max mem: 26157
Train: [20]  [ 100/6250]  eta: 0:33:47  lr: 0.000117  grad: 0.0806 (0.0893)  loss: 0.8312 (0.8342)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [20]  [ 200/6250]  eta: 0:30:15  lr: 0.000117  grad: 0.0792 (0.0852)  loss: 0.8215 (0.8316)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [20]  [ 300/6250]  eta: 0:28:44  lr: 0.000117  grad: 0.0688 (0.0837)  loss: 0.8266 (0.8276)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [20]  [ 400/6250]  eta: 0:27:44  lr: 0.000117  grad: 0.0753 (0.0821)  loss: 0.8251 (0.8266)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [20]  [ 500/6250]  eta: 0:26:59  lr: 0.000117  grad: 0.0783 (0.0813)  loss: 0.8232 (0.8261)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [20]  [ 600/6250]  eta: 0:26:20  lr: 0.000117  grad: 0.0723 (0.0807)  loss: 0.8293 (0.8260)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [20]  [ 700/6250]  eta: 0:25:44  lr: 0.000117  grad: 0.0699 (0.0796)  loss: 0.8267 (0.8263)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [20]  [ 800/6250]  eta: 0:25:10  lr: 0.000117  grad: 0.0698 (0.0791)  loss: 0.8284 (0.8262)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [20]  [ 900/6250]  eta: 0:24:38  lr: 0.000117  grad: 0.0747 (0.0785)  loss: 0.8234 (0.8260)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [20]  [1000/6250]  eta: 0:24:07  lr: 0.000117  grad: 0.0679 (0.0778)  loss: 0.8256 (0.8258)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [20]  [1100/6250]  eta: 0:23:36  lr: 0.000117  grad: 0.0660 (0.0780)  loss: 0.8328 (0.8259)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [20]  [1200/6250]  eta: 0:23:08  lr: 0.000117  grad: 0.0746 (0.0778)  loss: 0.8245 (0.8257)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [20]  [1300/6250]  eta: 0:22:39  lr: 0.000117  grad: 0.0719 (0.0775)  loss: 0.8262 (0.8257)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [20]  [1400/6250]  eta: 0:22:09  lr: 0.000117  grad: 0.0729 (0.0772)  loss: 0.8299 (0.8257)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [20]  [1500/6250]  eta: 0:21:41  lr: 0.000117  grad: 0.0711 (0.0770)  loss: 0.8272 (0.8255)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [20]  [1600/6250]  eta: 0:21:13  lr: 0.000117  grad: 0.0760 (0.0769)  loss: 0.8302 (0.8256)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [20]  [1700/6250]  eta: 0:20:44  lr: 0.000117  grad: 0.0732 (0.0768)  loss: 0.8248 (0.8256)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [20]  [1800/6250]  eta: 0:20:16  lr: 0.000117  grad: 0.0731 (0.0767)  loss: 0.8251 (0.8255)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [20]  [1900/6250]  eta: 0:19:47  lr: 0.000117  grad: 0.0750 (0.0766)  loss: 0.8234 (0.8255)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [20]  [2000/6250]  eta: 0:19:19  lr: 0.000117  grad: 0.0748 (0.0765)  loss: 0.8186 (0.8253)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [20]  [2100/6250]  eta: 0:18:56  lr: 0.000117  grad: 0.0724 (0.0764)  loss: 0.8186 (0.8252)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [20]  [2200/6250]  eta: 0:18:28  lr: 0.000117  grad: 0.0731 (0.0764)  loss: 0.8236 (0.8251)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [20]  [2300/6250]  eta: 0:18:05  lr: 0.000117  grad: 0.0713 (0.0763)  loss: 0.8227 (0.8251)  time: 0.2790  data: 0.0002  max mem: 26157
Train: [20]  [2400/6250]  eta: 0:17:36  lr: 0.000117  grad: 0.0784 (0.0763)  loss: 0.8255 (0.8252)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [20]  [2500/6250]  eta: 0:17:08  lr: 0.000117  grad: 0.0762 (0.0763)  loss: 0.8196 (0.8251)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [20]  [2600/6250]  eta: 0:16:40  lr: 0.000117  grad: 0.0781 (0.0764)  loss: 0.8223 (0.8250)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [20]  [2700/6250]  eta: 0:16:12  lr: 0.000117  grad: 0.0763 (0.0763)  loss: 0.8220 (0.8250)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [20]  [2800/6250]  eta: 0:15:44  lr: 0.000117  grad: 0.0779 (0.0763)  loss: 0.8209 (0.8250)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [20]  [2900/6250]  eta: 0:15:16  lr: 0.000117  grad: 0.0728 (0.0763)  loss: 0.8266 (0.8250)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [20]  [3000/6250]  eta: 0:14:48  lr: 0.000117  grad: 0.0716 (0.0764)  loss: 0.8277 (0.8250)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [20]  [3100/6250]  eta: 0:14:20  lr: 0.000117  grad: 0.0741 (0.0764)  loss: 0.8233 (0.8251)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [20]  [3200/6250]  eta: 0:13:53  lr: 0.000117  grad: 0.0737 (0.0764)  loss: 0.8262 (0.8251)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [20]  [3300/6250]  eta: 0:13:25  lr: 0.000117  grad: 0.0743 (0.0764)  loss: 0.8270 (0.8251)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [20]  [3400/6250]  eta: 0:12:58  lr: 0.000117  grad: 0.0816 (0.0765)  loss: 0.8235 (0.8251)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [20]  [3500/6250]  eta: 0:12:30  lr: 0.000117  grad: 0.0737 (0.0765)  loss: 0.8246 (0.8252)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [20]  [3600/6250]  eta: 0:12:03  lr: 0.000117  grad: 0.0733 (0.0765)  loss: 0.8212 (0.8252)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [20]  [3700/6250]  eta: 0:11:35  lr: 0.000117  grad: 0.0729 (0.0765)  loss: 0.8259 (0.8251)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [20]  [3800/6250]  eta: 0:11:08  lr: 0.000117  grad: 0.0730 (0.0766)  loss: 0.8305 (0.8251)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [20]  [3900/6250]  eta: 0:10:40  lr: 0.000117  grad: 0.0829 (0.0767)  loss: 0.8250 (0.8251)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [20]  [4000/6250]  eta: 0:10:14  lr: 0.000117  grad: 0.0746 (0.0767)  loss: 0.8307 (0.8251)  time: 0.3309  data: 0.0574  max mem: 26157
Train: [20]  [4100/6250]  eta: 0:09:49  lr: 0.000117  grad: 0.0755 (0.0768)  loss: 0.8228 (0.8251)  time: 0.5051  data: 0.2219  max mem: 26157
Train: [20]  [4200/6250]  eta: 0:09:21  lr: 0.000117  grad: 0.0746 (0.0768)  loss: 0.8269 (0.8251)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [20]  [4300/6250]  eta: 0:08:53  lr: 0.000117  grad: 0.0837 (0.0769)  loss: 0.8238 (0.8251)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [20]  [4400/6250]  eta: 0:08:26  lr: 0.000117  grad: 0.0751 (0.0769)  loss: 0.8262 (0.8250)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [20]  [4500/6250]  eta: 0:08:02  lr: 0.000117  grad: 0.0805 (0.0770)  loss: 0.8226 (0.8251)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [20]  [4600/6250]  eta: 0:07:34  lr: 0.000117  grad: 0.0747 (0.0771)  loss: 0.8293 (0.8251)  time: 0.2717  data: 0.0003  max mem: 26157
Train: [20]  [4700/6250]  eta: 0:07:07  lr: 0.000117  grad: 0.0796 (0.0772)  loss: 0.8257 (0.8251)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [20]  [4800/6250]  eta: 0:06:39  lr: 0.000117  grad: 0.0730 (0.0772)  loss: 0.8274 (0.8251)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [20]  [4900/6250]  eta: 0:06:11  lr: 0.000117  grad: 0.0736 (0.0773)  loss: 0.8275 (0.8250)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [20]  [5000/6250]  eta: 0:05:43  lr: 0.000117  grad: 0.0769 (0.0773)  loss: 0.8263 (0.8251)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [20]  [5100/6250]  eta: 0:05:16  lr: 0.000117  grad: 0.0856 (0.0774)  loss: 0.8190 (0.8250)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [20]  [5200/6250]  eta: 0:04:48  lr: 0.000117  grad: 0.0748 (0.0774)  loss: 0.8192 (0.8251)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [20]  [5300/6250]  eta: 0:04:21  lr: 0.000117  grad: 0.0719 (0.0774)  loss: 0.8228 (0.8251)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [20]  [5400/6250]  eta: 0:03:53  lr: 0.000117  grad: 0.0731 (0.0774)  loss: 0.8281 (0.8251)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [20]  [5500/6250]  eta: 0:03:25  lr: 0.000117  grad: 0.0727 (0.0774)  loss: 0.8276 (0.8252)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [20]  [5600/6250]  eta: 0:02:58  lr: 0.000117  grad: 0.0783 (0.0775)  loss: 0.8218 (0.8251)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [20]  [5700/6250]  eta: 0:02:30  lr: 0.000117  grad: 0.0743 (0.0775)  loss: 0.8279 (0.8251)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [20]  [5800/6250]  eta: 0:02:03  lr: 0.000117  grad: 0.0707 (0.0775)  loss: 0.8277 (0.8251)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [20]  [5900/6250]  eta: 0:01:36  lr: 0.000117  grad: 0.0777 (0.0775)  loss: 0.8223 (0.8251)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [20]  [6000/6250]  eta: 0:01:08  lr: 0.000116  grad: 0.0708 (0.0775)  loss: 0.8291 (0.8251)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [20]  [6100/6250]  eta: 0:00:41  lr: 0.000116  grad: 0.0766 (0.0775)  loss: 0.8219 (0.8251)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [20]  [6200/6250]  eta: 0:00:13  lr: 0.000116  grad: 0.0772 (0.0776)  loss: 0.8251 (0.8251)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [20]  [6249/6250]  eta: 0:00:00  lr: 0.000116  grad: 0.0736 (0.0776)  loss: 0.8258 (0.8251)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [20] Total time: 0:28:41 (0.2754 s / it)
Averaged stats: lr: 0.000116  grad: 0.0736 (0.0776)  loss: 0.8258 (0.8251)
Eval (hcp-train-subset): [20]  [ 0/62]  eta: 0:05:04  loss: 0.8491 (0.8491)  time: 4.9137  data: 4.8179  max mem: 26157
Eval (hcp-train-subset): [20]  [61/62]  eta: 0:00:00  loss: 0.8351 (0.8378)  time: 0.1372  data: 0.0541  max mem: 26157
Eval (hcp-train-subset): [20] Total time: 0:00:13 (0.2117 s / it)
Averaged stats (hcp-train-subset): loss: 0.8351 (0.8378)
Making plots (hcp-train-subset): example=30
Eval (hcp-val): [20]  [ 0/62]  eta: 0:04:19  loss: 0.8302 (0.8302)  time: 4.1809  data: 4.0887  max mem: 26157
Eval (hcp-val): [20]  [61/62]  eta: 0:00:00  loss: 0.8317 (0.8335)  time: 0.1327  data: 0.0499  max mem: 26157
Eval (hcp-val): [20] Total time: 0:00:12 (0.2074 s / it)
Averaged stats (hcp-val): loss: 0.8317 (0.8335)
Making plots (hcp-val): example=25
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [21]  [   0/6250]  eta: 11:09:42  lr: 0.000116  grad: 0.1017 (0.1017)  loss: 0.8293 (0.8293)  time: 6.4292  data: 6.1545  max mem: 26157
Train: [21]  [ 100/6250]  eta: 0:33:57  lr: 0.000116  grad: 0.0803 (0.0876)  loss: 0.8289 (0.8340)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [21]  [ 200/6250]  eta: 0:30:19  lr: 0.000116  grad: 0.0782 (0.0836)  loss: 0.8242 (0.8280)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [21]  [ 300/6250]  eta: 0:28:47  lr: 0.000116  grad: 0.0720 (0.0814)  loss: 0.8237 (0.8263)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [21]  [ 400/6250]  eta: 0:27:47  lr: 0.000116  grad: 0.0707 (0.0797)  loss: 0.8240 (0.8261)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [21]  [ 500/6250]  eta: 0:27:02  lr: 0.000116  grad: 0.0700 (0.0790)  loss: 0.8266 (0.8261)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [21]  [ 600/6250]  eta: 0:26:21  lr: 0.000116  grad: 0.0726 (0.0779)  loss: 0.8249 (0.8260)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [21]  [ 700/6250]  eta: 0:25:45  lr: 0.000116  grad: 0.0767 (0.0776)  loss: 0.8219 (0.8255)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [21]  [ 800/6250]  eta: 0:25:11  lr: 0.000116  grad: 0.0771 (0.0773)  loss: 0.8302 (0.8256)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [21]  [ 900/6250]  eta: 0:24:38  lr: 0.000116  grad: 0.0692 (0.0771)  loss: 0.8260 (0.8255)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [21]  [1000/6250]  eta: 0:24:07  lr: 0.000116  grad: 0.0737 (0.0769)  loss: 0.8244 (0.8255)  time: 0.2682  data: 0.0003  max mem: 26157
Train: [21]  [1100/6250]  eta: 0:23:36  lr: 0.000116  grad: 0.0752 (0.0769)  loss: 0.8159 (0.8253)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [21]  [1200/6250]  eta: 0:23:06  lr: 0.000116  grad: 0.0775 (0.0769)  loss: 0.8203 (0.8252)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [21]  [1300/6250]  eta: 0:22:37  lr: 0.000116  grad: 0.0712 (0.0769)  loss: 0.8166 (0.8247)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [21]  [1400/6250]  eta: 0:22:08  lr: 0.000116  grad: 0.0732 (0.0769)  loss: 0.8142 (0.8246)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [21]  [1500/6250]  eta: 0:21:39  lr: 0.000116  grad: 0.0733 (0.0771)  loss: 0.8145 (0.8243)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [21]  [1600/6250]  eta: 0:21:16  lr: 0.000116  grad: 0.0712 (0.0772)  loss: 0.8199 (0.8241)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [21]  [1700/6250]  eta: 0:20:54  lr: 0.000116  grad: 0.0761 (0.0771)  loss: 0.8251 (0.8240)  time: 0.2714  data: 0.0020  max mem: 26157
Train: [21]  [1800/6250]  eta: 0:20:26  lr: 0.000116  grad: 0.0752 (0.0771)  loss: 0.8186 (0.8239)  time: 0.2767  data: 0.0002  max mem: 26157
Train: [21]  [1900/6250]  eta: 0:20:05  lr: 0.000116  grad: 0.0711 (0.0770)  loss: 0.8231 (0.8238)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [21]  [2000/6250]  eta: 0:19:36  lr: 0.000116  grad: 0.0710 (0.0768)  loss: 0.8219 (0.8238)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [21]  [2100/6250]  eta: 0:19:07  lr: 0.000116  grad: 0.0771 (0.0768)  loss: 0.8267 (0.8238)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [21]  [2200/6250]  eta: 0:18:38  lr: 0.000116  grad: 0.0768 (0.0767)  loss: 0.8180 (0.8238)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [21]  [2300/6250]  eta: 0:18:09  lr: 0.000116  grad: 0.0720 (0.0767)  loss: 0.8214 (0.8238)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [21]  [2400/6250]  eta: 0:17:41  lr: 0.000116  grad: 0.0738 (0.0768)  loss: 0.8295 (0.8237)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [21]  [2500/6250]  eta: 0:17:12  lr: 0.000116  grad: 0.0816 (0.0769)  loss: 0.8268 (0.8237)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [21]  [2600/6250]  eta: 0:16:44  lr: 0.000116  grad: 0.0778 (0.0769)  loss: 0.8148 (0.8236)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [21]  [2700/6250]  eta: 0:16:16  lr: 0.000116  grad: 0.0724 (0.0769)  loss: 0.8223 (0.8236)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [21]  [2800/6250]  eta: 0:15:49  lr: 0.000116  grad: 0.0799 (0.0770)  loss: 0.8273 (0.8235)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [21]  [2900/6250]  eta: 0:15:21  lr: 0.000116  grad: 0.0740 (0.0770)  loss: 0.8245 (0.8235)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [21]  [3000/6250]  eta: 0:14:53  lr: 0.000116  grad: 0.0808 (0.0771)  loss: 0.8187 (0.8234)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [21]  [3100/6250]  eta: 0:14:26  lr: 0.000116  grad: 0.0761 (0.0771)  loss: 0.8204 (0.8234)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [21]  [3200/6250]  eta: 0:13:58  lr: 0.000116  grad: 0.0769 (0.0772)  loss: 0.8225 (0.8234)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [21]  [3300/6250]  eta: 0:13:30  lr: 0.000116  grad: 0.0802 (0.0773)  loss: 0.8231 (0.8233)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [21]  [3400/6250]  eta: 0:13:02  lr: 0.000116  grad: 0.0760 (0.0774)  loss: 0.8205 (0.8233)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [21]  [3500/6250]  eta: 0:12:34  lr: 0.000116  grad: 0.0718 (0.0774)  loss: 0.8290 (0.8233)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [21]  [3600/6250]  eta: 0:12:06  lr: 0.000116  grad: 0.0681 (0.0776)  loss: 0.8190 (0.8233)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [21]  [3700/6250]  eta: 0:11:39  lr: 0.000116  grad: 0.0738 (0.0776)  loss: 0.8287 (0.8233)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [21]  [3800/6250]  eta: 0:11:11  lr: 0.000116  grad: 0.0805 (0.0776)  loss: 0.8172 (0.8233)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [21]  [3900/6250]  eta: 0:10:43  lr: 0.000116  grad: 0.0776 (0.0777)  loss: 0.8263 (0.8233)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [21]  [4000/6250]  eta: 0:10:16  lr: 0.000116  grad: 0.0781 (0.0777)  loss: 0.8237 (0.8234)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [21]  [4100/6250]  eta: 0:09:48  lr: 0.000116  grad: 0.0758 (0.0777)  loss: 0.8225 (0.8234)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [21]  [4200/6250]  eta: 0:09:20  lr: 0.000116  grad: 0.0757 (0.0777)  loss: 0.8297 (0.8234)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [21]  [4300/6250]  eta: 0:08:53  lr: 0.000116  grad: 0.0755 (0.0777)  loss: 0.8296 (0.8235)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [21]  [4400/6250]  eta: 0:08:25  lr: 0.000116  grad: 0.0698 (0.0777)  loss: 0.8241 (0.8235)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [21]  [4500/6250]  eta: 0:07:58  lr: 0.000116  grad: 0.0733 (0.0777)  loss: 0.8251 (0.8236)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [21]  [4600/6250]  eta: 0:07:30  lr: 0.000116  grad: 0.0727 (0.0776)  loss: 0.8298 (0.8236)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [21]  [4700/6250]  eta: 0:07:03  lr: 0.000116  grad: 0.0803 (0.0777)  loss: 0.8265 (0.8236)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [21]  [4800/6250]  eta: 0:06:35  lr: 0.000116  grad: 0.0784 (0.0777)  loss: 0.8182 (0.8236)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [21]  [4900/6250]  eta: 0:06:08  lr: 0.000116  grad: 0.0732 (0.0778)  loss: 0.8234 (0.8236)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [21]  [5000/6250]  eta: 0:05:41  lr: 0.000116  grad: 0.0762 (0.0778)  loss: 0.8233 (0.8236)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [21]  [5100/6250]  eta: 0:05:13  lr: 0.000116  grad: 0.0711 (0.0778)  loss: 0.8247 (0.8236)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [21]  [5200/6250]  eta: 0:04:46  lr: 0.000116  grad: 0.0696 (0.0778)  loss: 0.8211 (0.8236)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [21]  [5300/6250]  eta: 0:04:19  lr: 0.000116  grad: 0.0742 (0.0777)  loss: 0.8237 (0.8236)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [21]  [5400/6250]  eta: 0:03:51  lr: 0.000116  grad: 0.0752 (0.0778)  loss: 0.8169 (0.8236)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [21]  [5500/6250]  eta: 0:03:24  lr: 0.000116  grad: 0.0750 (0.0778)  loss: 0.8257 (0.8236)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [21]  [5600/6250]  eta: 0:02:57  lr: 0.000115  grad: 0.0700 (0.0778)  loss: 0.8293 (0.8236)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [21]  [5700/6250]  eta: 0:02:30  lr: 0.000115  grad: 0.0801 (0.0777)  loss: 0.8245 (0.8237)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [21]  [5800/6250]  eta: 0:02:02  lr: 0.000115  grad: 0.0722 (0.0778)  loss: 0.8270 (0.8237)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [21]  [5900/6250]  eta: 0:01:35  lr: 0.000115  grad: 0.0762 (0.0778)  loss: 0.8233 (0.8237)  time: 0.2764  data: 0.0002  max mem: 26157
Train: [21]  [6000/6250]  eta: 0:01:08  lr: 0.000115  grad: 0.0780 (0.0777)  loss: 0.8277 (0.8237)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [21]  [6100/6250]  eta: 0:00:40  lr: 0.000115  grad: 0.0725 (0.0777)  loss: 0.8313 (0.8238)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [21]  [6200/6250]  eta: 0:00:13  lr: 0.000115  grad: 0.0769 (0.0777)  loss: 0.8253 (0.8239)  time: 0.3031  data: 0.0316  max mem: 26157
Train: [21]  [6249/6250]  eta: 0:00:00  lr: 0.000115  grad: 0.0748 (0.0777)  loss: 0.8287 (0.8239)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [21] Total time: 0:28:35 (0.2745 s / it)
Averaged stats: lr: 0.000115  grad: 0.0748 (0.0777)  loss: 0.8287 (0.8239)
Eval (hcp-train-subset): [21]  [ 0/62]  eta: 0:04:16  loss: 0.8487 (0.8487)  time: 4.1352  data: 4.0223  max mem: 26157
Eval (hcp-train-subset): [21]  [61/62]  eta: 0:00:00  loss: 0.8368 (0.8374)  time: 0.1228  data: 0.0382  max mem: 26157
Eval (hcp-train-subset): [21] Total time: 0:00:13 (0.2130 s / it)
Averaged stats (hcp-train-subset): loss: 0.8368 (0.8374)
Making plots (hcp-train-subset): example=56
Eval (hcp-val): [21]  [ 0/62]  eta: 0:04:38  loss: 0.8295 (0.8295)  time: 4.4986  data: 4.4147  max mem: 26157
Eval (hcp-val): [21]  [61/62]  eta: 0:00:00  loss: 0.8312 (0.8327)  time: 0.1360  data: 0.0531  max mem: 26157
Eval (hcp-val): [21] Total time: 0:00:13 (0.2138 s / it)
Averaged stats (hcp-val): loss: 0.8312 (0.8327)
Making plots (hcp-val): example=26
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [22]  [   0/6250]  eta: 8:44:51  lr: 0.000115  grad: 0.1219 (0.1219)  loss: 0.8218 (0.8218)  time: 5.0387  data: 4.6837  max mem: 26157
Train: [22]  [ 100/6250]  eta: 0:33:05  lr: 0.000115  grad: 0.0818 (0.0894)  loss: 0.8220 (0.8275)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [22]  [ 200/6250]  eta: 0:29:54  lr: 0.000115  grad: 0.0847 (0.0897)  loss: 0.8074 (0.8214)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [22]  [ 300/6250]  eta: 0:28:33  lr: 0.000115  grad: 0.0809 (0.0882)  loss: 0.8163 (0.8188)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [22]  [ 400/6250]  eta: 0:27:39  lr: 0.000115  grad: 0.0794 (0.0870)  loss: 0.8178 (0.8182)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [22]  [ 500/6250]  eta: 0:26:54  lr: 0.000115  grad: 0.0784 (0.0863)  loss: 0.8188 (0.8180)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [22]  [ 600/6250]  eta: 0:26:15  lr: 0.000115  grad: 0.0746 (0.0860)  loss: 0.8209 (0.8176)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [22]  [ 700/6250]  eta: 0:25:41  lr: 0.000115  grad: 0.0745 (0.0848)  loss: 0.8232 (0.8181)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [22]  [ 800/6250]  eta: 0:25:07  lr: 0.000115  grad: 0.0733 (0.0840)  loss: 0.8270 (0.8183)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [22]  [ 900/6250]  eta: 0:24:36  lr: 0.000115  grad: 0.0746 (0.0834)  loss: 0.8203 (0.8188)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [22]  [1000/6250]  eta: 0:24:17  lr: 0.000115  grad: 0.0785 (0.0827)  loss: 0.8232 (0.8191)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [22]  [1100/6250]  eta: 0:23:45  lr: 0.000115  grad: 0.0738 (0.0824)  loss: 0.8231 (0.8193)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [22]  [1200/6250]  eta: 0:23:16  lr: 0.000115  grad: 0.0790 (0.0823)  loss: 0.8185 (0.8193)  time: 0.2701  data: 0.0003  max mem: 26157
Train: [22]  [1300/6250]  eta: 0:22:46  lr: 0.000115  grad: 0.0752 (0.0821)  loss: 0.8163 (0.8193)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [22]  [1400/6250]  eta: 0:22:16  lr: 0.000115  grad: 0.0767 (0.0820)  loss: 0.8219 (0.8193)  time: 0.2710  data: 0.0015  max mem: 26157
Train: [22]  [1500/6250]  eta: 0:21:46  lr: 0.000115  grad: 0.0749 (0.0817)  loss: 0.8223 (0.8194)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [22]  [1600/6250]  eta: 0:21:18  lr: 0.000115  grad: 0.0870 (0.0817)  loss: 0.8208 (0.8195)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [22]  [1700/6250]  eta: 0:20:49  lr: 0.000115  grad: 0.0793 (0.0816)  loss: 0.8183 (0.8194)  time: 0.2756  data: 0.0002  max mem: 26157
Train: [22]  [1800/6250]  eta: 0:20:20  lr: 0.000115  grad: 0.0818 (0.0815)  loss: 0.8195 (0.8194)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [22]  [1900/6250]  eta: 0:19:52  lr: 0.000115  grad: 0.0788 (0.0814)  loss: 0.8200 (0.8195)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [22]  [2000/6250]  eta: 0:19:23  lr: 0.000115  grad: 0.0780 (0.0814)  loss: 0.8177 (0.8193)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [22]  [2100/6250]  eta: 0:18:55  lr: 0.000115  grad: 0.0804 (0.0814)  loss: 0.8172 (0.8194)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [22]  [2200/6250]  eta: 0:18:27  lr: 0.000115  grad: 0.0829 (0.0815)  loss: 0.8170 (0.8193)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [22]  [2300/6250]  eta: 0:17:59  lr: 0.000115  grad: 0.0795 (0.0814)  loss: 0.8185 (0.8193)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [22]  [2400/6250]  eta: 0:17:31  lr: 0.000115  grad: 0.0770 (0.0814)  loss: 0.8201 (0.8193)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [22]  [2500/6250]  eta: 0:17:03  lr: 0.000115  grad: 0.0777 (0.0813)  loss: 0.8170 (0.8193)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [22]  [2600/6250]  eta: 0:16:35  lr: 0.000115  grad: 0.0810 (0.0814)  loss: 0.8221 (0.8192)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [22]  [2700/6250]  eta: 0:16:07  lr: 0.000115  grad: 0.0827 (0.0815)  loss: 0.8131 (0.8191)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [22]  [2800/6250]  eta: 0:15:40  lr: 0.000115  grad: 0.0815 (0.0818)  loss: 0.8151 (0.8191)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [22]  [2900/6250]  eta: 0:15:12  lr: 0.000115  grad: 0.0822 (0.0819)  loss: 0.8127 (0.8190)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [22]  [3000/6250]  eta: 0:14:45  lr: 0.000115  grad: 0.0776 (0.0819)  loss: 0.8224 (0.8190)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [22]  [3100/6250]  eta: 0:14:23  lr: 0.000115  grad: 0.0809 (0.0820)  loss: 0.8133 (0.8190)  time: 0.5392  data: 0.2647  max mem: 26157
Train: [22]  [3200/6250]  eta: 0:13:55  lr: 0.000115  grad: 0.0798 (0.0820)  loss: 0.8182 (0.8190)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [22]  [3300/6250]  eta: 0:13:27  lr: 0.000115  grad: 0.0767 (0.0820)  loss: 0.8239 (0.8190)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [22]  [3400/6250]  eta: 0:12:59  lr: 0.000115  grad: 0.0839 (0.0820)  loss: 0.8211 (0.8190)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [22]  [3500/6250]  eta: 0:12:32  lr: 0.000115  grad: 0.0763 (0.0820)  loss: 0.8247 (0.8191)  time: 0.2694  data: 0.0001  max mem: 26157
Train: [22]  [3600/6250]  eta: 0:12:05  lr: 0.000115  grad: 0.0815 (0.0820)  loss: 0.8266 (0.8191)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [22]  [3700/6250]  eta: 0:11:37  lr: 0.000115  grad: 0.0769 (0.0820)  loss: 0.8255 (0.8192)  time: 0.2757  data: 0.0003  max mem: 26157
Train: [22]  [3800/6250]  eta: 0:11:10  lr: 0.000115  grad: 0.0782 (0.0820)  loss: 0.8175 (0.8192)  time: 0.2760  data: 0.0002  max mem: 26157
Train: [22]  [3900/6250]  eta: 0:10:43  lr: 0.000115  grad: 0.0811 (0.0820)  loss: 0.8168 (0.8193)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [22]  [4000/6250]  eta: 0:10:15  lr: 0.000115  grad: 0.0791 (0.0819)  loss: 0.8253 (0.8193)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [22]  [4100/6250]  eta: 0:09:47  lr: 0.000115  grad: 0.0818 (0.0819)  loss: 0.8238 (0.8194)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [22]  [4200/6250]  eta: 0:09:20  lr: 0.000115  grad: 0.0774 (0.0819)  loss: 0.8286 (0.8195)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [22]  [4300/6250]  eta: 0:08:53  lr: 0.000115  grad: 0.0826 (0.0818)  loss: 0.8180 (0.8195)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [22]  [4400/6250]  eta: 0:08:25  lr: 0.000115  grad: 0.0730 (0.0818)  loss: 0.8219 (0.8196)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [22]  [4500/6250]  eta: 0:07:58  lr: 0.000115  grad: 0.0767 (0.0818)  loss: 0.8211 (0.8196)  time: 0.2707  data: 0.0003  max mem: 26157
Train: [22]  [4600/6250]  eta: 0:07:30  lr: 0.000115  grad: 0.0766 (0.0817)  loss: 0.8262 (0.8197)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [22]  [4700/6250]  eta: 0:07:03  lr: 0.000115  grad: 0.0818 (0.0817)  loss: 0.8233 (0.8198)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [22]  [4800/6250]  eta: 0:06:35  lr: 0.000115  grad: 0.0780 (0.0817)  loss: 0.8178 (0.8198)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [22]  [4900/6250]  eta: 0:06:08  lr: 0.000114  grad: 0.0791 (0.0817)  loss: 0.8205 (0.8199)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [22]  [5000/6250]  eta: 0:05:41  lr: 0.000114  grad: 0.0833 (0.0817)  loss: 0.8250 (0.8199)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [22]  [5100/6250]  eta: 0:05:13  lr: 0.000114  grad: 0.0746 (0.0817)  loss: 0.8237 (0.8199)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [22]  [5200/6250]  eta: 0:04:46  lr: 0.000114  grad: 0.0777 (0.0817)  loss: 0.8236 (0.8200)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [22]  [5300/6250]  eta: 0:04:19  lr: 0.000114  grad: 0.0789 (0.0816)  loss: 0.8230 (0.8200)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [22]  [5400/6250]  eta: 0:03:51  lr: 0.000114  grad: 0.0780 (0.0816)  loss: 0.8242 (0.8200)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [22]  [5500/6250]  eta: 0:03:24  lr: 0.000114  grad: 0.0780 (0.0815)  loss: 0.8252 (0.8201)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [22]  [5600/6250]  eta: 0:02:57  lr: 0.000114  grad: 0.0803 (0.0815)  loss: 0.8283 (0.8201)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [22]  [5700/6250]  eta: 0:02:29  lr: 0.000114  grad: 0.0774 (0.0815)  loss: 0.8240 (0.8202)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [22]  [5800/6250]  eta: 0:02:02  lr: 0.000114  grad: 0.0781 (0.0815)  loss: 0.8258 (0.8202)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [22]  [5900/6250]  eta: 0:01:35  lr: 0.000114  grad: 0.0841 (0.0815)  loss: 0.8198 (0.8202)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [22]  [6000/6250]  eta: 0:01:08  lr: 0.000114  grad: 0.0785 (0.0815)  loss: 0.8289 (0.8203)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [22]  [6100/6250]  eta: 0:00:40  lr: 0.000114  grad: 0.0817 (0.0815)  loss: 0.8231 (0.8203)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [22]  [6200/6250]  eta: 0:00:13  lr: 0.000114  grad: 0.0799 (0.0816)  loss: 0.8287 (0.8203)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [22]  [6249/6250]  eta: 0:00:00  lr: 0.000114  grad: 0.0784 (0.0816)  loss: 0.8207 (0.8204)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [22] Total time: 0:28:27 (0.2732 s / it)
Averaged stats: lr: 0.000114  grad: 0.0784 (0.0816)  loss: 0.8207 (0.8204)
Eval (hcp-train-subset): [22]  [ 0/62]  eta: 0:03:16  loss: 0.8445 (0.8445)  time: 3.1659  data: 3.0461  max mem: 26157
Eval (hcp-train-subset): [22]  [61/62]  eta: 0:00:00  loss: 0.8338 (0.8350)  time: 0.1231  data: 0.0400  max mem: 26157
Eval (hcp-train-subset): [22] Total time: 0:00:12 (0.2012 s / it)
Averaged stats (hcp-train-subset): loss: 0.8338 (0.8350)
Making plots (hcp-train-subset): example=9
Eval (hcp-val): [22]  [ 0/62]  eta: 0:05:13  loss: 0.8274 (0.8274)  time: 5.0510  data: 4.9670  max mem: 26157
Eval (hcp-val): [22]  [61/62]  eta: 0:00:00  loss: 0.8325 (0.8323)  time: 0.1348  data: 0.0497  max mem: 26157
Eval (hcp-val): [22] Total time: 0:00:12 (0.2034 s / it)
Averaged stats (hcp-val): loss: 0.8325 (0.8323)
Making plots (hcp-val): example=54
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [23]  [   0/6250]  eta: 7:50:31  lr: 0.000114  grad: 0.1248 (0.1248)  loss: 0.8269 (0.8269)  time: 4.5170  data: 4.1735  max mem: 26157
Train: [23]  [ 100/6250]  eta: 0:33:47  lr: 0.000114  grad: 0.0710 (0.0924)  loss: 0.8299 (0.8291)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [23]  [ 200/6250]  eta: 0:30:19  lr: 0.000114  grad: 0.0755 (0.0860)  loss: 0.8205 (0.8276)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [23]  [ 300/6250]  eta: 0:29:38  lr: 0.000114  grad: 0.0741 (0.0832)  loss: 0.8220 (0.8281)  time: 0.3734  data: 0.1018  max mem: 26157
Train: [23]  [ 400/6250]  eta: 0:28:29  lr: 0.000114  grad: 0.0683 (0.0814)  loss: 0.8313 (0.8283)  time: 0.2749  data: 0.0004  max mem: 26157
Train: [23]  [ 500/6250]  eta: 0:27:45  lr: 0.000114  grad: 0.0743 (0.0813)  loss: 0.8273 (0.8280)  time: 0.3132  data: 0.0351  max mem: 26157
Train: [23]  [ 600/6250]  eta: 0:27:00  lr: 0.000114  grad: 0.0744 (0.0810)  loss: 0.8233 (0.8275)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [23]  [ 700/6250]  eta: 0:26:20  lr: 0.000114  grad: 0.0787 (0.0806)  loss: 0.8268 (0.8273)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [23]  [ 800/6250]  eta: 0:25:43  lr: 0.000114  grad: 0.0789 (0.0803)  loss: 0.8254 (0.8274)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [23]  [ 900/6250]  eta: 0:25:09  lr: 0.000114  grad: 0.0750 (0.0799)  loss: 0.8223 (0.8273)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [23]  [1000/6250]  eta: 0:24:58  lr: 0.000114  grad: 0.0728 (0.0795)  loss: 0.8203 (0.8270)  time: 0.4858  data: 0.2161  max mem: 26157
Train: [23]  [1100/6250]  eta: 0:24:22  lr: 0.000114  grad: 0.0712 (0.0791)  loss: 0.8323 (0.8269)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [23]  [1200/6250]  eta: 0:23:48  lr: 0.000114  grad: 0.0782 (0.0793)  loss: 0.8248 (0.8265)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [23]  [1300/6250]  eta: 0:23:15  lr: 0.000114  grad: 0.0740 (0.0792)  loss: 0.8236 (0.8263)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [23]  [1400/6250]  eta: 0:22:42  lr: 0.000114  grad: 0.0748 (0.0792)  loss: 0.8225 (0.8260)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [23]  [1500/6250]  eta: 0:22:11  lr: 0.000114  grad: 0.0789 (0.0795)  loss: 0.8172 (0.8258)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [23]  [1600/6250]  eta: 0:21:40  lr: 0.000114  grad: 0.0849 (0.0797)  loss: 0.8217 (0.8255)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [23]  [1700/6250]  eta: 0:21:30  lr: 0.000114  grad: 0.0748 (0.0797)  loss: 0.8229 (0.8252)  time: 0.6389  data: 0.3603  max mem: 26157
Train: [23]  [1800/6250]  eta: 0:20:58  lr: 0.000114  grad: 0.0805 (0.0798)  loss: 0.8235 (0.8248)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [23]  [1900/6250]  eta: 0:20:27  lr: 0.000114  grad: 0.0769 (0.0798)  loss: 0.8199 (0.8245)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [23]  [2000/6250]  eta: 0:20:03  lr: 0.000114  grad: 0.0848 (0.0799)  loss: 0.8170 (0.8243)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [23]  [2100/6250]  eta: 0:19:32  lr: 0.000114  grad: 0.0760 (0.0800)  loss: 0.8198 (0.8241)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [23]  [2200/6250]  eta: 0:19:01  lr: 0.000114  grad: 0.0798 (0.0800)  loss: 0.8207 (0.8240)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [23]  [2300/6250]  eta: 0:18:31  lr: 0.000114  grad: 0.0770 (0.0801)  loss: 0.8214 (0.8239)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [23]  [2400/6250]  eta: 0:18:01  lr: 0.000114  grad: 0.0750 (0.0800)  loss: 0.8232 (0.8238)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [23]  [2500/6250]  eta: 0:17:32  lr: 0.000114  grad: 0.0771 (0.0800)  loss: 0.8220 (0.8238)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [23]  [2600/6250]  eta: 0:17:02  lr: 0.000114  grad: 0.0805 (0.0800)  loss: 0.8247 (0.8237)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [23]  [2700/6250]  eta: 0:16:33  lr: 0.000114  grad: 0.0769 (0.0800)  loss: 0.8224 (0.8238)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [23]  [2800/6250]  eta: 0:16:04  lr: 0.000114  grad: 0.0770 (0.0800)  loss: 0.8250 (0.8238)  time: 0.2737  data: 0.0003  max mem: 26157
Train: [23]  [2900/6250]  eta: 0:15:35  lr: 0.000114  grad: 0.0797 (0.0800)  loss: 0.8236 (0.8238)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [23]  [3000/6250]  eta: 0:15:20  lr: 0.000114  grad: 0.0829 (0.0799)  loss: 0.8209 (0.8238)  time: 0.2785  data: 0.0002  max mem: 26157
Train: [23]  [3100/6250]  eta: 0:14:51  lr: 0.000114  grad: 0.0805 (0.0799)  loss: 0.8230 (0.8238)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [23]  [3200/6250]  eta: 0:14:21  lr: 0.000114  grad: 0.0750 (0.0799)  loss: 0.8249 (0.8237)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [23]  [3300/6250]  eta: 0:13:52  lr: 0.000114  grad: 0.0849 (0.0799)  loss: 0.8192 (0.8236)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [23]  [3400/6250]  eta: 0:13:23  lr: 0.000114  grad: 0.0775 (0.0799)  loss: 0.8220 (0.8235)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [23]  [3500/6250]  eta: 0:12:53  lr: 0.000114  grad: 0.0803 (0.0799)  loss: 0.8145 (0.8234)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [23]  [3600/6250]  eta: 0:12:24  lr: 0.000114  grad: 0.0833 (0.0801)  loss: 0.8227 (0.8234)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [23]  [3700/6250]  eta: 0:11:55  lr: 0.000114  grad: 0.0812 (0.0801)  loss: 0.8218 (0.8233)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [23]  [3800/6250]  eta: 0:11:27  lr: 0.000114  grad: 0.0715 (0.0802)  loss: 0.8246 (0.8233)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [23]  [3900/6250]  eta: 0:10:58  lr: 0.000114  grad: 0.0784 (0.0802)  loss: 0.8216 (0.8233)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [23]  [4000/6250]  eta: 0:10:29  lr: 0.000113  grad: 0.0776 (0.0802)  loss: 0.8213 (0.8232)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [23]  [4100/6250]  eta: 0:10:01  lr: 0.000113  grad: 0.0780 (0.0802)  loss: 0.8236 (0.8232)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [23]  [4200/6250]  eta: 0:09:32  lr: 0.000113  grad: 0.0793 (0.0802)  loss: 0.8170 (0.8231)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [23]  [4300/6250]  eta: 0:09:04  lr: 0.000113  grad: 0.0754 (0.0802)  loss: 0.8207 (0.8231)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [23]  [4400/6250]  eta: 0:08:36  lr: 0.000113  grad: 0.0794 (0.0802)  loss: 0.8128 (0.8230)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [23]  [4500/6250]  eta: 0:08:07  lr: 0.000113  grad: 0.0791 (0.0802)  loss: 0.8210 (0.8229)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [23]  [4600/6250]  eta: 0:07:39  lr: 0.000113  grad: 0.0775 (0.0802)  loss: 0.8236 (0.8229)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [23]  [4700/6250]  eta: 0:07:11  lr: 0.000113  grad: 0.0796 (0.0803)  loss: 0.8196 (0.8229)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [23]  [4800/6250]  eta: 0:06:43  lr: 0.000113  grad: 0.0769 (0.0803)  loss: 0.8220 (0.8228)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [23]  [4900/6250]  eta: 0:06:15  lr: 0.000113  grad: 0.0805 (0.0803)  loss: 0.8162 (0.8228)  time: 0.2729  data: 0.0003  max mem: 26157
Train: [23]  [5000/6250]  eta: 0:05:47  lr: 0.000113  grad: 0.0758 (0.0802)  loss: 0.8256 (0.8229)  time: 0.2756  data: 0.0002  max mem: 26157
Train: [23]  [5100/6250]  eta: 0:05:19  lr: 0.000113  grad: 0.0785 (0.0803)  loss: 0.8247 (0.8228)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [23]  [5200/6250]  eta: 0:04:51  lr: 0.000113  grad: 0.0753 (0.0803)  loss: 0.8215 (0.8228)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [23]  [5300/6250]  eta: 0:04:23  lr: 0.000113  grad: 0.0773 (0.0803)  loss: 0.8237 (0.8227)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [23]  [5400/6250]  eta: 0:03:56  lr: 0.000113  grad: 0.0804 (0.0803)  loss: 0.8232 (0.8227)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [23]  [5500/6250]  eta: 0:03:28  lr: 0.000113  grad: 0.0837 (0.0803)  loss: 0.8165 (0.8227)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [23]  [5600/6250]  eta: 0:03:00  lr: 0.000113  grad: 0.0799 (0.0804)  loss: 0.8219 (0.8227)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [23]  [5700/6250]  eta: 0:02:32  lr: 0.000113  grad: 0.0778 (0.0804)  loss: 0.8246 (0.8227)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [23]  [5800/6250]  eta: 0:02:04  lr: 0.000113  grad: 0.0791 (0.0804)  loss: 0.8172 (0.8227)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [23]  [5900/6250]  eta: 0:01:37  lr: 0.000113  grad: 0.0819 (0.0804)  loss: 0.8211 (0.8226)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [23]  [6000/6250]  eta: 0:01:09  lr: 0.000113  grad: 0.0758 (0.0804)  loss: 0.8218 (0.8226)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [23]  [6100/6250]  eta: 0:00:41  lr: 0.000113  grad: 0.0769 (0.0804)  loss: 0.8209 (0.8225)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [23]  [6200/6250]  eta: 0:00:13  lr: 0.000113  grad: 0.0794 (0.0805)  loss: 0.8235 (0.8225)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [23]  [6249/6250]  eta: 0:00:00  lr: 0.000113  grad: 0.0728 (0.0805)  loss: 0.8197 (0.8226)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [23] Total time: 0:28:56 (0.2779 s / it)
Averaged stats: lr: 0.000113  grad: 0.0728 (0.0805)  loss: 0.8197 (0.8226)
Eval (hcp-train-subset): [23]  [ 0/62]  eta: 0:03:50  loss: 0.8437 (0.8437)  time: 3.7174  data: 3.6079  max mem: 26157
Eval (hcp-train-subset): [23]  [61/62]  eta: 0:00:00  loss: 0.8335 (0.8355)  time: 0.1103  data: 0.0259  max mem: 26157
Eval (hcp-train-subset): [23] Total time: 0:00:12 (0.2052 s / it)
Averaged stats (hcp-train-subset): loss: 0.8335 (0.8355)
Making plots (hcp-train-subset): example=44
Eval (hcp-val): [23]  [ 0/62]  eta: 0:03:48  loss: 0.8257 (0.8257)  time: 3.6868  data: 3.5643  max mem: 26157
Eval (hcp-val): [23]  [61/62]  eta: 0:00:00  loss: 0.8310 (0.8321)  time: 0.1289  data: 0.0455  max mem: 26157
Eval (hcp-val): [23] Total time: 0:00:13 (0.2216 s / it)
Averaged stats (hcp-val): loss: 0.8310 (0.8321)
Making plots (hcp-val): example=23
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [24]  [   0/6250]  eta: 8:35:38  lr: 0.000113  grad: 0.0654 (0.0654)  loss: 0.8552 (0.8552)  time: 4.9502  data: 4.5488  max mem: 26157
Train: [24]  [ 100/6250]  eta: 0:33:10  lr: 0.000113  grad: 0.0742 (0.0898)  loss: 0.8305 (0.8348)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [24]  [ 200/6250]  eta: 0:29:56  lr: 0.000113  grad: 0.0733 (0.0846)  loss: 0.8323 (0.8302)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [24]  [ 300/6250]  eta: 0:28:31  lr: 0.000113  grad: 0.0758 (0.0832)  loss: 0.8229 (0.8272)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [24]  [ 400/6250]  eta: 0:27:37  lr: 0.000113  grad: 0.0778 (0.0817)  loss: 0.8140 (0.8252)  time: 0.2751  data: 0.0002  max mem: 26157
Train: [24]  [ 500/6250]  eta: 0:26:53  lr: 0.000113  grad: 0.0736 (0.0807)  loss: 0.8221 (0.8244)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [24]  [ 600/6250]  eta: 0:26:15  lr: 0.000113  grad: 0.0710 (0.0801)  loss: 0.8251 (0.8240)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [24]  [ 700/6250]  eta: 0:25:41  lr: 0.000113  grad: 0.0738 (0.0801)  loss: 0.8184 (0.8236)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [24]  [ 800/6250]  eta: 0:25:09  lr: 0.000113  grad: 0.0832 (0.0803)  loss: 0.8191 (0.8233)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [24]  [ 900/6250]  eta: 0:24:38  lr: 0.000113  grad: 0.0807 (0.0801)  loss: 0.8212 (0.8232)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [24]  [1000/6250]  eta: 0:24:07  lr: 0.000113  grad: 0.0757 (0.0798)  loss: 0.8283 (0.8233)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [24]  [1100/6250]  eta: 0:23:37  lr: 0.000113  grad: 0.0736 (0.0797)  loss: 0.8253 (0.8234)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [24]  [1200/6250]  eta: 0:23:06  lr: 0.000113  grad: 0.0763 (0.0796)  loss: 0.8192 (0.8234)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [24]  [1300/6250]  eta: 0:22:37  lr: 0.000113  grad: 0.0770 (0.0795)  loss: 0.8255 (0.8234)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [24]  [1400/6250]  eta: 0:22:08  lr: 0.000113  grad: 0.0787 (0.0793)  loss: 0.8285 (0.8235)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [24]  [1500/6250]  eta: 0:21:39  lr: 0.000113  grad: 0.0744 (0.0793)  loss: 0.8219 (0.8234)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [24]  [1600/6250]  eta: 0:21:13  lr: 0.000113  grad: 0.0802 (0.0795)  loss: 0.8226 (0.8233)  time: 0.3084  data: 0.0384  max mem: 26157
Train: [24]  [1700/6250]  eta: 0:20:44  lr: 0.000113  grad: 0.0776 (0.0799)  loss: 0.8225 (0.8232)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [24]  [1800/6250]  eta: 0:20:33  lr: 0.000113  grad: 0.0772 (0.0799)  loss: 0.8210 (0.8229)  time: 0.6115  data: 0.3320  max mem: 26157
Train: [24]  [1900/6250]  eta: 0:20:04  lr: 0.000113  grad: 0.0791 (0.0799)  loss: 0.8144 (0.8228)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [24]  [2000/6250]  eta: 0:19:35  lr: 0.000113  grad: 0.0746 (0.0799)  loss: 0.8229 (0.8226)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [24]  [2100/6250]  eta: 0:19:11  lr: 0.000113  grad: 0.0749 (0.0799)  loss: 0.8219 (0.8225)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [24]  [2200/6250]  eta: 0:19:00  lr: 0.000113  grad: 0.0782 (0.0799)  loss: 0.8245 (0.8225)  time: 0.7434  data: 0.4708  max mem: 26157
Train: [24]  [2300/6250]  eta: 0:18:30  lr: 0.000113  grad: 0.0775 (0.0800)  loss: 0.8236 (0.8225)  time: 0.2739  data: 0.0002  max mem: 26157
Train: [24]  [2400/6250]  eta: 0:18:01  lr: 0.000113  grad: 0.0758 (0.0800)  loss: 0.8203 (0.8225)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [24]  [2500/6250]  eta: 0:17:31  lr: 0.000113  grad: 0.0769 (0.0801)  loss: 0.8172 (0.8225)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [24]  [2600/6250]  eta: 0:17:02  lr: 0.000113  grad: 0.0791 (0.0802)  loss: 0.8182 (0.8224)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [24]  [2700/6250]  eta: 0:16:32  lr: 0.000113  grad: 0.0784 (0.0802)  loss: 0.8225 (0.8224)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [24]  [2800/6250]  eta: 0:16:05  lr: 0.000113  grad: 0.0815 (0.0804)  loss: 0.8211 (0.8224)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [24]  [2900/6250]  eta: 0:15:36  lr: 0.000112  grad: 0.0804 (0.0805)  loss: 0.8204 (0.8224)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [24]  [3000/6250]  eta: 0:15:07  lr: 0.000112  grad: 0.0812 (0.0807)  loss: 0.8254 (0.8224)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [24]  [3100/6250]  eta: 0:14:38  lr: 0.000112  grad: 0.0823 (0.0811)  loss: 0.8217 (0.8223)  time: 0.2726  data: 0.0003  max mem: 26157
Train: [24]  [3200/6250]  eta: 0:14:09  lr: 0.000112  grad: 0.0809 (0.0812)  loss: 0.8209 (0.8222)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [24]  [3300/6250]  eta: 0:13:41  lr: 0.000112  grad: 0.0823 (0.0813)  loss: 0.8195 (0.8222)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [24]  [3400/6250]  eta: 0:13:12  lr: 0.000112  grad: 0.0840 (0.0813)  loss: 0.8223 (0.8222)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [24]  [3500/6250]  eta: 0:12:44  lr: 0.000112  grad: 0.0815 (0.0814)  loss: 0.8188 (0.8222)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [24]  [3600/6250]  eta: 0:12:15  lr: 0.000112  grad: 0.0803 (0.0814)  loss: 0.8152 (0.8221)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [24]  [3700/6250]  eta: 0:11:47  lr: 0.000112  grad: 0.0805 (0.0814)  loss: 0.8201 (0.8221)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [24]  [3800/6250]  eta: 0:11:19  lr: 0.000112  grad: 0.0812 (0.0815)  loss: 0.8199 (0.8220)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [24]  [3900/6250]  eta: 0:10:51  lr: 0.000112  grad: 0.0789 (0.0815)  loss: 0.8169 (0.8219)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [24]  [4000/6250]  eta: 0:10:22  lr: 0.000112  grad: 0.0777 (0.0815)  loss: 0.8217 (0.8218)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [24]  [4100/6250]  eta: 0:09:59  lr: 0.000112  grad: 0.0787 (0.0815)  loss: 0.8233 (0.8218)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [24]  [4200/6250]  eta: 0:09:34  lr: 0.000112  grad: 0.0850 (0.0815)  loss: 0.8192 (0.8219)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [24]  [4300/6250]  eta: 0:09:05  lr: 0.000112  grad: 0.0788 (0.0815)  loss: 0.8195 (0.8219)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [24]  [4400/6250]  eta: 0:08:39  lr: 0.000112  grad: 0.0783 (0.0816)  loss: 0.8177 (0.8218)  time: 0.5457  data: 0.2696  max mem: 26157
Train: [24]  [4500/6250]  eta: 0:08:11  lr: 0.000112  grad: 0.0799 (0.0816)  loss: 0.8170 (0.8218)  time: 0.2794  data: 0.0002  max mem: 26157
Train: [24]  [4600/6250]  eta: 0:07:46  lr: 0.000112  grad: 0.0814 (0.0816)  loss: 0.8157 (0.8217)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [24]  [4700/6250]  eta: 0:07:20  lr: 0.000112  grad: 0.0778 (0.0816)  loss: 0.8208 (0.8217)  time: 0.5257  data: 0.2530  max mem: 26157
Train: [24]  [4800/6250]  eta: 0:06:51  lr: 0.000112  grad: 0.0799 (0.0817)  loss: 0.8222 (0.8217)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [24]  [4900/6250]  eta: 0:06:22  lr: 0.000112  grad: 0.0761 (0.0817)  loss: 0.8191 (0.8217)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [24]  [5000/6250]  eta: 0:05:53  lr: 0.000112  grad: 0.0762 (0.0816)  loss: 0.8253 (0.8216)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [24]  [5100/6250]  eta: 0:05:25  lr: 0.000112  grad: 0.0830 (0.0817)  loss: 0.8186 (0.8216)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [24]  [5200/6250]  eta: 0:04:56  lr: 0.000112  grad: 0.0833 (0.0817)  loss: 0.8218 (0.8216)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [24]  [5300/6250]  eta: 0:04:28  lr: 0.000112  grad: 0.0811 (0.0817)  loss: 0.8139 (0.8216)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [24]  [5400/6250]  eta: 0:03:59  lr: 0.000112  grad: 0.0810 (0.0817)  loss: 0.8223 (0.8216)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [24]  [5500/6250]  eta: 0:03:31  lr: 0.000112  grad: 0.0776 (0.0818)  loss: 0.8211 (0.8216)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [24]  [5600/6250]  eta: 0:03:03  lr: 0.000112  grad: 0.0849 (0.0818)  loss: 0.8216 (0.8215)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [24]  [5700/6250]  eta: 0:02:34  lr: 0.000112  grad: 0.0804 (0.0818)  loss: 0.8186 (0.8215)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [24]  [5800/6250]  eta: 0:02:06  lr: 0.000112  grad: 0.0776 (0.0819)  loss: 0.8224 (0.8214)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [24]  [5900/6250]  eta: 0:01:38  lr: 0.000112  grad: 0.0830 (0.0820)  loss: 0.8170 (0.8213)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [24]  [6000/6250]  eta: 0:01:10  lr: 0.000112  grad: 0.0804 (0.0820)  loss: 0.8157 (0.8213)  time: 0.3295  data: 0.0582  max mem: 26157
Train: [24]  [6100/6250]  eta: 0:00:42  lr: 0.000112  grad: 0.0858 (0.0821)  loss: 0.8145 (0.8212)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [24]  [6200/6250]  eta: 0:00:14  lr: 0.000112  grad: 0.0800 (0.0821)  loss: 0.8178 (0.8212)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [24]  [6249/6250]  eta: 0:00:00  lr: 0.000112  grad: 0.0796 (0.0821)  loss: 0.8265 (0.8212)  time: 0.2757  data: 0.0002  max mem: 26157
Train: [24] Total time: 0:29:27 (0.2828 s / it)
Averaged stats: lr: 0.000112  grad: 0.0796 (0.0821)  loss: 0.8265 (0.8212)
Eval (hcp-train-subset): [24]  [ 0/62]  eta: 0:04:43  loss: 0.8434 (0.8434)  time: 4.5731  data: 4.4613  max mem: 26157
Eval (hcp-train-subset): [24]  [61/62]  eta: 0:00:00  loss: 0.8328 (0.8344)  time: 0.1488  data: 0.0655  max mem: 26157
Eval (hcp-train-subset): [24] Total time: 0:00:13 (0.2242 s / it)
Averaged stats (hcp-train-subset): loss: 0.8328 (0.8344)
Making plots (hcp-train-subset): example=38
Eval (hcp-val): [24]  [ 0/62]  eta: 0:03:48  loss: 0.8293 (0.8293)  time: 3.6911  data: 3.5789  max mem: 26157
Eval (hcp-val): [24]  [61/62]  eta: 0:00:00  loss: 0.8304 (0.8314)  time: 0.1499  data: 0.0669  max mem: 26157
Eval (hcp-val): [24] Total time: 0:00:13 (0.2211 s / it)
Averaged stats (hcp-val): loss: 0.8304 (0.8314)
Making plots (hcp-val): example=30
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [25]  [   0/6250]  eta: 11:11:37  lr: 0.000112  grad: 0.0759 (0.0759)  loss: 0.8620 (0.8620)  time: 6.4476  data: 6.1699  max mem: 26157
Train: [25]  [ 100/6250]  eta: 0:34:10  lr: 0.000112  grad: 0.0700 (0.0826)  loss: 0.8362 (0.8372)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [25]  [ 200/6250]  eta: 0:30:27  lr: 0.000112  grad: 0.0809 (0.0815)  loss: 0.8305 (0.8355)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [25]  [ 300/6250]  eta: 0:28:54  lr: 0.000112  grad: 0.0768 (0.0813)  loss: 0.8252 (0.8322)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [25]  [ 400/6250]  eta: 0:27:55  lr: 0.000112  grad: 0.0816 (0.0821)  loss: 0.8188 (0.8292)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [25]  [ 500/6250]  eta: 0:27:12  lr: 0.000112  grad: 0.0810 (0.0828)  loss: 0.8221 (0.8268)  time: 0.2766  data: 0.0002  max mem: 26157
Train: [25]  [ 600/6250]  eta: 0:26:47  lr: 0.000112  grad: 0.0772 (0.0826)  loss: 0.8213 (0.8256)  time: 0.3537  data: 0.0816  max mem: 26157
Train: [25]  [ 700/6250]  eta: 0:26:08  lr: 0.000112  grad: 0.0797 (0.0822)  loss: 0.8236 (0.8249)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [25]  [ 800/6250]  eta: 0:25:31  lr: 0.000112  grad: 0.0747 (0.0819)  loss: 0.8255 (0.8245)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [25]  [ 900/6250]  eta: 0:24:59  lr: 0.000112  grad: 0.0791 (0.0817)  loss: 0.8188 (0.8240)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [25]  [1000/6250]  eta: 0:24:26  lr: 0.000112  grad: 0.0812 (0.0818)  loss: 0.8218 (0.8236)  time: 0.2744  data: 0.0002  max mem: 26157
Train: [25]  [1100/6250]  eta: 0:23:54  lr: 0.000112  grad: 0.0753 (0.0817)  loss: 0.8219 (0.8234)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [25]  [1200/6250]  eta: 0:23:22  lr: 0.000112  grad: 0.0774 (0.0816)  loss: 0.8160 (0.8230)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [25]  [1300/6250]  eta: 0:22:51  lr: 0.000112  grad: 0.0797 (0.0819)  loss: 0.8161 (0.8225)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [25]  [1400/6250]  eta: 0:22:21  lr: 0.000112  grad: 0.0757 (0.0822)  loss: 0.8166 (0.8220)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [25]  [1500/6250]  eta: 0:21:50  lr: 0.000112  grad: 0.0772 (0.0822)  loss: 0.8213 (0.8217)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [25]  [1600/6250]  eta: 0:21:21  lr: 0.000111  grad: 0.0811 (0.0824)  loss: 0.8126 (0.8214)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [25]  [1700/6250]  eta: 0:20:52  lr: 0.000111  grad: 0.0847 (0.0826)  loss: 0.8186 (0.8210)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [25]  [1800/6250]  eta: 0:20:23  lr: 0.000111  grad: 0.0850 (0.0826)  loss: 0.8215 (0.8207)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [25]  [1900/6250]  eta: 0:19:54  lr: 0.000111  grad: 0.0790 (0.0827)  loss: 0.8160 (0.8203)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [25]  [2000/6250]  eta: 0:19:25  lr: 0.000111  grad: 0.0815 (0.0827)  loss: 0.8110 (0.8199)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [25]  [2100/6250]  eta: 0:18:57  lr: 0.000111  grad: 0.0831 (0.0829)  loss: 0.8173 (0.8196)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [25]  [2200/6250]  eta: 0:18:29  lr: 0.000111  grad: 0.0770 (0.0830)  loss: 0.8105 (0.8193)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [25]  [2300/6250]  eta: 0:18:01  lr: 0.000111  grad: 0.0800 (0.0832)  loss: 0.8162 (0.8192)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [25]  [2400/6250]  eta: 0:17:33  lr: 0.000111  grad: 0.0743 (0.0831)  loss: 0.8150 (0.8190)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [25]  [2500/6250]  eta: 0:17:05  lr: 0.000111  grad: 0.0822 (0.0831)  loss: 0.8154 (0.8188)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [25]  [2600/6250]  eta: 0:16:37  lr: 0.000111  grad: 0.0806 (0.0831)  loss: 0.8122 (0.8188)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [25]  [2700/6250]  eta: 0:16:09  lr: 0.000111  grad: 0.0819 (0.0832)  loss: 0.8100 (0.8186)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [25]  [2800/6250]  eta: 0:15:41  lr: 0.000111  grad: 0.0857 (0.0832)  loss: 0.8132 (0.8186)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [25]  [2900/6250]  eta: 0:15:13  lr: 0.000111  grad: 0.0834 (0.0833)  loss: 0.8145 (0.8185)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [25]  [3000/6250]  eta: 0:14:46  lr: 0.000111  grad: 0.0845 (0.0835)  loss: 0.8178 (0.8184)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [25]  [3100/6250]  eta: 0:14:18  lr: 0.000111  grad: 0.0806 (0.0836)  loss: 0.8181 (0.8184)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [25]  [3200/6250]  eta: 0:13:51  lr: 0.000111  grad: 0.0845 (0.0837)  loss: 0.8146 (0.8183)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [25]  [3300/6250]  eta: 0:13:24  lr: 0.000111  grad: 0.0790 (0.0836)  loss: 0.8213 (0.8183)  time: 0.2901  data: 0.0003  max mem: 26157
Train: [25]  [3400/6250]  eta: 0:12:57  lr: 0.000111  grad: 0.0847 (0.0836)  loss: 0.8178 (0.8183)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [25]  [3500/6250]  eta: 0:12:29  lr: 0.000111  grad: 0.0800 (0.0837)  loss: 0.8193 (0.8183)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [25]  [3600/6250]  eta: 0:12:02  lr: 0.000111  grad: 0.0850 (0.0838)  loss: 0.8120 (0.8182)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [25]  [3700/6250]  eta: 0:11:36  lr: 0.000111  grad: 0.0810 (0.0838)  loss: 0.8217 (0.8182)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [25]  [3800/6250]  eta: 0:11:08  lr: 0.000111  grad: 0.0797 (0.0838)  loss: 0.8206 (0.8181)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [25]  [3900/6250]  eta: 0:10:42  lr: 0.000111  grad: 0.0805 (0.0837)  loss: 0.8206 (0.8181)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [25]  [4000/6250]  eta: 0:10:14  lr: 0.000111  grad: 0.0806 (0.0837)  loss: 0.8156 (0.8180)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [25]  [4100/6250]  eta: 0:09:47  lr: 0.000111  grad: 0.0838 (0.0838)  loss: 0.8150 (0.8180)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [25]  [4200/6250]  eta: 0:09:19  lr: 0.000111  grad: 0.0880 (0.0839)  loss: 0.8182 (0.8179)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [25]  [4300/6250]  eta: 0:08:52  lr: 0.000111  grad: 0.0886 (0.0840)  loss: 0.8145 (0.8178)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [25]  [4400/6250]  eta: 0:08:25  lr: 0.000111  grad: 0.0816 (0.0840)  loss: 0.8143 (0.8177)  time: 0.2781  data: 0.0002  max mem: 26157
Train: [25]  [4500/6250]  eta: 0:07:57  lr: 0.000111  grad: 0.0829 (0.0842)  loss: 0.8118 (0.8176)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [25]  [4600/6250]  eta: 0:07:30  lr: 0.000111  grad: 0.0822 (0.0842)  loss: 0.8184 (0.8175)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [25]  [4700/6250]  eta: 0:07:03  lr: 0.000111  grad: 0.0828 (0.0843)  loss: 0.8179 (0.8174)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [25]  [4800/6250]  eta: 0:06:35  lr: 0.000111  grad: 0.0813 (0.0843)  loss: 0.8195 (0.8174)  time: 0.2768  data: 0.0002  max mem: 26157
Train: [25]  [4900/6250]  eta: 0:06:08  lr: 0.000111  grad: 0.0877 (0.0843)  loss: 0.8191 (0.8174)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [25]  [5000/6250]  eta: 0:05:40  lr: 0.000111  grad: 0.0809 (0.0843)  loss: 0.8220 (0.8174)  time: 0.2701  data: 0.0001  max mem: 26157
Train: [25]  [5100/6250]  eta: 0:05:13  lr: 0.000111  grad: 0.0786 (0.0843)  loss: 0.8115 (0.8174)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [25]  [5200/6250]  eta: 0:04:46  lr: 0.000111  grad: 0.0787 (0.0843)  loss: 0.8270 (0.8175)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [25]  [5300/6250]  eta: 0:04:18  lr: 0.000111  grad: 0.0795 (0.0843)  loss: 0.8199 (0.8175)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [25]  [5400/6250]  eta: 0:03:51  lr: 0.000111  grad: 0.0845 (0.0843)  loss: 0.8174 (0.8176)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [25]  [5500/6250]  eta: 0:03:24  lr: 0.000111  grad: 0.0876 (0.0844)  loss: 0.8195 (0.8176)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [25]  [5600/6250]  eta: 0:02:57  lr: 0.000111  grad: 0.0818 (0.0844)  loss: 0.8193 (0.8176)  time: 0.2695  data: 0.0003  max mem: 26157
Train: [25]  [5700/6250]  eta: 0:02:29  lr: 0.000111  grad: 0.0803 (0.0844)  loss: 0.8206 (0.8177)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [25]  [5800/6250]  eta: 0:02:02  lr: 0.000111  grad: 0.0807 (0.0844)  loss: 0.8192 (0.8177)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [25]  [5900/6250]  eta: 0:01:35  lr: 0.000111  grad: 0.0805 (0.0844)  loss: 0.8207 (0.8177)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [25]  [6000/6250]  eta: 0:01:08  lr: 0.000111  grad: 0.0836 (0.0843)  loss: 0.8220 (0.8178)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [25]  [6100/6250]  eta: 0:00:40  lr: 0.000111  grad: 0.0874 (0.0844)  loss: 0.8196 (0.8178)  time: 0.4334  data: 0.1577  max mem: 26157
Train: [25]  [6200/6250]  eta: 0:00:13  lr: 0.000111  grad: 0.0859 (0.0844)  loss: 0.8114 (0.8178)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [25]  [6249/6250]  eta: 0:00:00  lr: 0.000111  grad: 0.0800 (0.0844)  loss: 0.8155 (0.8178)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [25] Total time: 0:28:31 (0.2738 s / it)
Averaged stats: lr: 0.000111  grad: 0.0800 (0.0844)  loss: 0.8155 (0.8178)
Eval (hcp-train-subset): [25]  [ 0/62]  eta: 0:04:51  loss: 0.8455 (0.8455)  time: 4.7002  data: 4.6158  max mem: 26157
Eval (hcp-train-subset): [25]  [61/62]  eta: 0:00:00  loss: 0.8333 (0.8339)  time: 0.1266  data: 0.0435  max mem: 26157
Eval (hcp-train-subset): [25] Total time: 0:00:12 (0.2049 s / it)
Averaged stats (hcp-train-subset): loss: 0.8333 (0.8339)
Making plots (hcp-train-subset): example=35
Eval (hcp-val): [25]  [ 0/62]  eta: 0:03:24  loss: 0.8251 (0.8251)  time: 3.2909  data: 3.1760  max mem: 26157
Eval (hcp-val): [25]  [61/62]  eta: 0:00:00  loss: 0.8300 (0.8306)  time: 0.1326  data: 0.0500  max mem: 26157
Eval (hcp-val): [25] Total time: 0:00:12 (0.2066 s / it)
Averaged stats (hcp-val): loss: 0.8300 (0.8306)
Making plots (hcp-val): example=26
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [26]  [   0/6250]  eta: 7:04:30  lr: 0.000111  grad: 0.0570 (0.0570)  loss: 0.8668 (0.8668)  time: 4.0753  data: 3.7289  max mem: 26157
Train: [26]  [ 100/6250]  eta: 0:34:25  lr: 0.000111  grad: 0.0733 (0.0855)  loss: 0.8353 (0.8347)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [26]  [ 200/6250]  eta: 0:30:31  lr: 0.000110  grad: 0.0735 (0.0875)  loss: 0.8247 (0.8286)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [26]  [ 300/6250]  eta: 0:28:55  lr: 0.000110  grad: 0.0908 (0.0884)  loss: 0.8255 (0.8256)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [26]  [ 400/6250]  eta: 0:27:54  lr: 0.000110  grad: 0.0742 (0.0868)  loss: 0.8291 (0.8249)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [26]  [ 500/6250]  eta: 0:27:09  lr: 0.000110  grad: 0.0809 (0.0859)  loss: 0.8175 (0.8242)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [26]  [ 600/6250]  eta: 0:26:28  lr: 0.000110  grad: 0.0747 (0.0847)  loss: 0.8197 (0.8243)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [26]  [ 700/6250]  eta: 0:25:50  lr: 0.000110  grad: 0.0766 (0.0840)  loss: 0.8178 (0.8241)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [26]  [ 800/6250]  eta: 0:25:16  lr: 0.000110  grad: 0.0750 (0.0836)  loss: 0.8262 (0.8239)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [26]  [ 900/6250]  eta: 0:24:44  lr: 0.000110  grad: 0.0777 (0.0832)  loss: 0.8272 (0.8237)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [26]  [1000/6250]  eta: 0:24:12  lr: 0.000110  grad: 0.0811 (0.0829)  loss: 0.8210 (0.8236)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [26]  [1100/6250]  eta: 0:23:41  lr: 0.000110  grad: 0.0796 (0.0828)  loss: 0.8192 (0.8234)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [26]  [1200/6250]  eta: 0:23:11  lr: 0.000110  grad: 0.0812 (0.0826)  loss: 0.8136 (0.8231)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [26]  [1300/6250]  eta: 0:22:41  lr: 0.000110  grad: 0.0837 (0.0827)  loss: 0.8170 (0.8228)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [26]  [1400/6250]  eta: 0:22:12  lr: 0.000110  grad: 0.0853 (0.0830)  loss: 0.8166 (0.8225)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [26]  [1500/6250]  eta: 0:21:43  lr: 0.000110  grad: 0.0805 (0.0830)  loss: 0.8153 (0.8221)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [26]  [1600/6250]  eta: 0:21:14  lr: 0.000110  grad: 0.0838 (0.0831)  loss: 0.8175 (0.8219)  time: 0.2697  data: 0.0001  max mem: 26157
Train: [26]  [1700/6250]  eta: 0:20:51  lr: 0.000110  grad: 0.0792 (0.0832)  loss: 0.8173 (0.8216)  time: 0.3850  data: 0.1146  max mem: 26157
Train: [26]  [1800/6250]  eta: 0:20:22  lr: 0.000110  grad: 0.0877 (0.0834)  loss: 0.8173 (0.8213)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [26]  [1900/6250]  eta: 0:19:54  lr: 0.000110  grad: 0.0860 (0.0835)  loss: 0.8248 (0.8212)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [26]  [2000/6250]  eta: 0:19:25  lr: 0.000110  grad: 0.0843 (0.0837)  loss: 0.8201 (0.8209)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [26]  [2100/6250]  eta: 0:18:57  lr: 0.000110  grad: 0.0822 (0.0837)  loss: 0.8198 (0.8208)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [26]  [2200/6250]  eta: 0:18:30  lr: 0.000110  grad: 0.0849 (0.0839)  loss: 0.8180 (0.8208)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [26]  [2300/6250]  eta: 0:18:02  lr: 0.000110  grad: 0.0793 (0.0839)  loss: 0.8255 (0.8208)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [26]  [2400/6250]  eta: 0:17:33  lr: 0.000110  grad: 0.0867 (0.0841)  loss: 0.8189 (0.8207)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [26]  [2500/6250]  eta: 0:17:05  lr: 0.000110  grad: 0.0816 (0.0841)  loss: 0.8243 (0.8207)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [26]  [2600/6250]  eta: 0:16:37  lr: 0.000110  grad: 0.0822 (0.0841)  loss: 0.8167 (0.8207)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [26]  [2700/6250]  eta: 0:16:09  lr: 0.000110  grad: 0.0802 (0.0842)  loss: 0.8237 (0.8207)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [26]  [2800/6250]  eta: 0:15:41  lr: 0.000110  grad: 0.0772 (0.0842)  loss: 0.8218 (0.8207)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [26]  [2900/6250]  eta: 0:15:14  lr: 0.000110  grad: 0.0859 (0.0842)  loss: 0.8229 (0.8207)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [26]  [3000/6250]  eta: 0:14:46  lr: 0.000110  grad: 0.0858 (0.0842)  loss: 0.8214 (0.8207)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [26]  [3100/6250]  eta: 0:14:19  lr: 0.000110  grad: 0.0786 (0.0841)  loss: 0.8188 (0.8208)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [26]  [3200/6250]  eta: 0:13:51  lr: 0.000110  grad: 0.0800 (0.0841)  loss: 0.8147 (0.8207)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [26]  [3300/6250]  eta: 0:13:24  lr: 0.000110  grad: 0.0769 (0.0840)  loss: 0.8213 (0.8208)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [26]  [3400/6250]  eta: 0:12:58  lr: 0.000110  grad: 0.0833 (0.0840)  loss: 0.8188 (0.8207)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [26]  [3500/6250]  eta: 0:12:31  lr: 0.000110  grad: 0.0873 (0.0840)  loss: 0.8120 (0.8206)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [26]  [3600/6250]  eta: 0:12:03  lr: 0.000110  grad: 0.0844 (0.0841)  loss: 0.8213 (0.8205)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [26]  [3700/6250]  eta: 0:11:36  lr: 0.000110  grad: 0.0805 (0.0841)  loss: 0.8197 (0.8205)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [26]  [3800/6250]  eta: 0:11:08  lr: 0.000110  grad: 0.0803 (0.0841)  loss: 0.8149 (0.8204)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [26]  [3900/6250]  eta: 0:10:40  lr: 0.000110  grad: 0.0824 (0.0840)  loss: 0.8234 (0.8203)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [26]  [4000/6250]  eta: 0:10:13  lr: 0.000110  grad: 0.0831 (0.0841)  loss: 0.8193 (0.8203)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [26]  [4100/6250]  eta: 0:09:46  lr: 0.000110  grad: 0.0778 (0.0841)  loss: 0.8243 (0.8204)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [26]  [4200/6250]  eta: 0:09:18  lr: 0.000110  grad: 0.0856 (0.0840)  loss: 0.8155 (0.8203)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [26]  [4300/6250]  eta: 0:08:51  lr: 0.000110  grad: 0.0832 (0.0841)  loss: 0.8213 (0.8203)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [26]  [4400/6250]  eta: 0:08:23  lr: 0.000110  grad: 0.0792 (0.0840)  loss: 0.8224 (0.8203)  time: 0.2696  data: 0.0003  max mem: 26157
Train: [26]  [4500/6250]  eta: 0:07:56  lr: 0.000110  grad: 0.0860 (0.0841)  loss: 0.8127 (0.8202)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [26]  [4600/6250]  eta: 0:07:29  lr: 0.000110  grad: 0.0835 (0.0841)  loss: 0.8185 (0.8202)  time: 0.2695  data: 0.0003  max mem: 26157
Train: [26]  [4700/6250]  eta: 0:07:01  lr: 0.000110  grad: 0.0799 (0.0841)  loss: 0.8155 (0.8201)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [26]  [4800/6250]  eta: 0:06:34  lr: 0.000109  grad: 0.0824 (0.0841)  loss: 0.8190 (0.8201)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [26]  [4900/6250]  eta: 0:06:07  lr: 0.000109  grad: 0.0777 (0.0840)  loss: 0.8185 (0.8201)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [26]  [5000/6250]  eta: 0:05:40  lr: 0.000109  grad: 0.0812 (0.0840)  loss: 0.8217 (0.8200)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [26]  [5100/6250]  eta: 0:05:12  lr: 0.000109  grad: 0.0802 (0.0840)  loss: 0.8145 (0.8200)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [26]  [5200/6250]  eta: 0:04:45  lr: 0.000109  grad: 0.0774 (0.0840)  loss: 0.8232 (0.8200)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [26]  [5300/6250]  eta: 0:04:18  lr: 0.000109  grad: 0.0847 (0.0840)  loss: 0.8187 (0.8199)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [26]  [5400/6250]  eta: 0:03:52  lr: 0.000109  grad: 0.0763 (0.0840)  loss: 0.8178 (0.8199)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [26]  [5500/6250]  eta: 0:03:24  lr: 0.000109  grad: 0.0826 (0.0840)  loss: 0.8118 (0.8199)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [26]  [5600/6250]  eta: 0:02:57  lr: 0.000109  grad: 0.0817 (0.0840)  loss: 0.8167 (0.8199)  time: 0.2964  data: 0.0261  max mem: 26157
Train: [26]  [5700/6250]  eta: 0:02:30  lr: 0.000109  grad: 0.0789 (0.0840)  loss: 0.8207 (0.8198)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [26]  [5800/6250]  eta: 0:02:02  lr: 0.000109  grad: 0.0800 (0.0839)  loss: 0.8178 (0.8198)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [26]  [5900/6250]  eta: 0:01:35  lr: 0.000109  grad: 0.0791 (0.0840)  loss: 0.8163 (0.8198)  time: 0.2739  data: 0.0002  max mem: 26157
Train: [26]  [6000/6250]  eta: 0:01:08  lr: 0.000109  grad: 0.0847 (0.0841)  loss: 0.8193 (0.8198)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [26]  [6100/6250]  eta: 0:00:40  lr: 0.000109  grad: 0.0826 (0.0842)  loss: 0.8165 (0.8197)  time: 0.2684  data: 0.0004  max mem: 26157
Train: [26]  [6200/6250]  eta: 0:00:13  lr: 0.000109  grad: 0.0862 (0.0843)  loss: 0.8149 (0.8197)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [26]  [6249/6250]  eta: 0:00:00  lr: 0.000109  grad: 0.0800 (0.0843)  loss: 0.8162 (0.8196)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [26] Total time: 0:28:32 (0.2740 s / it)
Averaged stats: lr: 0.000109  grad: 0.0800 (0.0843)  loss: 0.8162 (0.8196)
Eval (hcp-train-subset): [26]  [ 0/62]  eta: 0:03:08  loss: 0.8478 (0.8478)  time: 3.0373  data: 2.9162  max mem: 26157
Eval (hcp-train-subset): [26]  [61/62]  eta: 0:00:00  loss: 0.8337 (0.8338)  time: 0.1320  data: 0.0484  max mem: 26157
Eval (hcp-train-subset): [26] Total time: 0:00:12 (0.2011 s / it)
Averaged stats (hcp-train-subset): loss: 0.8337 (0.8338)
Making plots (hcp-train-subset): example=11
Eval (hcp-val): [26]  [ 0/62]  eta: 0:03:25  loss: 0.8255 (0.8255)  time: 3.3118  data: 3.2165  max mem: 26157
Eval (hcp-val): [26]  [61/62]  eta: 0:00:00  loss: 0.8299 (0.8307)  time: 0.1092  data: 0.0261  max mem: 26157
Eval (hcp-val): [26] Total time: 0:00:12 (0.1999 s / it)
Averaged stats (hcp-val): loss: 0.8299 (0.8307)
Making plots (hcp-val): example=7
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [27]  [   0/6250]  eta: 8:34:57  lr: 0.000109  grad: 0.2265 (0.2265)  loss: 0.8354 (0.8354)  time: 4.9436  data: 4.5906  max mem: 26157
Train: [27]  [ 100/6250]  eta: 0:33:21  lr: 0.000109  grad: 0.0912 (0.1068)  loss: 0.8168 (0.8177)  time: 0.2690  data: 0.0003  max mem: 26157
Train: [27]  [ 200/6250]  eta: 0:29:58  lr: 0.000109  grad: 0.0820 (0.0993)  loss: 0.8213 (0.8150)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [27]  [ 300/6250]  eta: 0:28:34  lr: 0.000109  grad: 0.0822 (0.0946)  loss: 0.8201 (0.8151)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [27]  [ 400/6250]  eta: 0:27:40  lr: 0.000109  grad: 0.0934 (0.0928)  loss: 0.8114 (0.8153)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [27]  [ 500/6250]  eta: 0:26:58  lr: 0.000109  grad: 0.0821 (0.0915)  loss: 0.8076 (0.8154)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [27]  [ 600/6250]  eta: 0:26:19  lr: 0.000109  grad: 0.0796 (0.0903)  loss: 0.8239 (0.8156)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [27]  [ 700/6250]  eta: 0:25:44  lr: 0.000109  grad: 0.0805 (0.0892)  loss: 0.8216 (0.8159)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [27]  [ 800/6250]  eta: 0:25:12  lr: 0.000109  grad: 0.0764 (0.0882)  loss: 0.8238 (0.8162)  time: 0.2773  data: 0.0002  max mem: 26157
Train: [27]  [ 900/6250]  eta: 0:24:52  lr: 0.000109  grad: 0.0813 (0.0874)  loss: 0.8175 (0.8166)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [27]  [1000/6250]  eta: 0:24:33  lr: 0.000109  grad: 0.0824 (0.0871)  loss: 0.8213 (0.8169)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [27]  [1100/6250]  eta: 0:24:00  lr: 0.000109  grad: 0.0765 (0.0865)  loss: 0.8196 (0.8172)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [27]  [1200/6250]  eta: 0:23:27  lr: 0.000109  grad: 0.0770 (0.0860)  loss: 0.8176 (0.8174)  time: 0.2689  data: 0.0003  max mem: 26157
Train: [27]  [1300/6250]  eta: 0:22:55  lr: 0.000109  grad: 0.0856 (0.0858)  loss: 0.8137 (0.8173)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [27]  [1400/6250]  eta: 0:22:25  lr: 0.000109  grad: 0.0849 (0.0858)  loss: 0.8122 (0.8172)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [27]  [1500/6250]  eta: 0:21:54  lr: 0.000109  grad: 0.0815 (0.0858)  loss: 0.8137 (0.8170)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [27]  [1600/6250]  eta: 0:21:24  lr: 0.000109  grad: 0.0822 (0.0859)  loss: 0.8136 (0.8167)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [27]  [1700/6250]  eta: 0:20:55  lr: 0.000109  grad: 0.0849 (0.0860)  loss: 0.8114 (0.8165)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [27]  [1800/6250]  eta: 0:20:26  lr: 0.000109  grad: 0.0863 (0.0862)  loss: 0.8121 (0.8162)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [27]  [1900/6250]  eta: 0:19:57  lr: 0.000109  grad: 0.0966 (0.0866)  loss: 0.8101 (0.8159)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [27]  [2000/6250]  eta: 0:19:28  lr: 0.000109  grad: 0.0871 (0.0866)  loss: 0.8145 (0.8156)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [27]  [2100/6250]  eta: 0:19:00  lr: 0.000109  grad: 0.0886 (0.0867)  loss: 0.8095 (0.8154)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [27]  [2200/6250]  eta: 0:18:32  lr: 0.000109  grad: 0.0852 (0.0869)  loss: 0.8099 (0.8151)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [27]  [2300/6250]  eta: 0:18:03  lr: 0.000109  grad: 0.0812 (0.0869)  loss: 0.8145 (0.8149)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [27]  [2400/6250]  eta: 0:17:35  lr: 0.000109  grad: 0.0883 (0.0871)  loss: 0.8067 (0.8148)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [27]  [2500/6250]  eta: 0:17:07  lr: 0.000109  grad: 0.0823 (0.0871)  loss: 0.8137 (0.8148)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [27]  [2600/6250]  eta: 0:16:39  lr: 0.000109  grad: 0.0871 (0.0872)  loss: 0.8137 (0.8146)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [27]  [2700/6250]  eta: 0:16:11  lr: 0.000109  grad: 0.0856 (0.0872)  loss: 0.8098 (0.8145)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [27]  [2800/6250]  eta: 0:15:44  lr: 0.000109  grad: 0.0864 (0.0873)  loss: 0.8113 (0.8144)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [27]  [2900/6250]  eta: 0:15:16  lr: 0.000109  grad: 0.0823 (0.0874)  loss: 0.8136 (0.8143)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [27]  [3000/6250]  eta: 0:14:48  lr: 0.000109  grad: 0.0845 (0.0873)  loss: 0.8132 (0.8143)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [27]  [3100/6250]  eta: 0:14:20  lr: 0.000108  grad: 0.0845 (0.0874)  loss: 0.8133 (0.8143)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [27]  [3200/6250]  eta: 0:13:53  lr: 0.000108  grad: 0.0869 (0.0874)  loss: 0.8150 (0.8143)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [27]  [3300/6250]  eta: 0:13:25  lr: 0.000108  grad: 0.0853 (0.0874)  loss: 0.8080 (0.8142)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [27]  [3400/6250]  eta: 0:12:57  lr: 0.000108  grad: 0.0879 (0.0874)  loss: 0.8144 (0.8142)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [27]  [3500/6250]  eta: 0:12:30  lr: 0.000108  grad: 0.0869 (0.0874)  loss: 0.8130 (0.8141)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [27]  [3600/6250]  eta: 0:12:02  lr: 0.000108  grad: 0.0888 (0.0874)  loss: 0.8127 (0.8141)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [27]  [3700/6250]  eta: 0:11:35  lr: 0.000108  grad: 0.0832 (0.0874)  loss: 0.8174 (0.8142)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [27]  [3800/6250]  eta: 0:11:10  lr: 0.000108  grad: 0.0855 (0.0874)  loss: 0.8166 (0.8142)  time: 0.3347  data: 0.0002  max mem: 26157
Train: [27]  [3900/6250]  eta: 0:10:43  lr: 0.000108  grad: 0.0867 (0.0873)  loss: 0.8136 (0.8143)  time: 0.2721  data: 0.0003  max mem: 26157
Train: [27]  [4000/6250]  eta: 0:10:15  lr: 0.000108  grad: 0.0845 (0.0874)  loss: 0.8195 (0.8144)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [27]  [4100/6250]  eta: 0:09:48  lr: 0.000108  grad: 0.0757 (0.0873)  loss: 0.8147 (0.8145)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [27]  [4200/6250]  eta: 0:09:20  lr: 0.000108  grad: 0.0852 (0.0872)  loss: 0.8155 (0.8145)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [27]  [4300/6250]  eta: 0:08:53  lr: 0.000108  grad: 0.0866 (0.0872)  loss: 0.8161 (0.8146)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [27]  [4400/6250]  eta: 0:08:25  lr: 0.000108  grad: 0.0916 (0.0872)  loss: 0.8102 (0.8146)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [27]  [4500/6250]  eta: 0:07:58  lr: 0.000108  grad: 0.0845 (0.0872)  loss: 0.8155 (0.8147)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [27]  [4600/6250]  eta: 0:07:30  lr: 0.000108  grad: 0.0933 (0.0872)  loss: 0.8151 (0.8147)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [27]  [4700/6250]  eta: 0:07:03  lr: 0.000108  grad: 0.0886 (0.0873)  loss: 0.8178 (0.8147)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [27]  [4800/6250]  eta: 0:06:35  lr: 0.000108  grad: 0.0858 (0.0874)  loss: 0.8168 (0.8148)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [27]  [4900/6250]  eta: 0:06:08  lr: 0.000108  grad: 0.0826 (0.0874)  loss: 0.8199 (0.8148)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [27]  [5000/6250]  eta: 0:05:41  lr: 0.000108  grad: 0.0929 (0.0874)  loss: 0.8084 (0.8148)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [27]  [5100/6250]  eta: 0:05:13  lr: 0.000108  grad: 0.0866 (0.0875)  loss: 0.8100 (0.8148)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [27]  [5200/6250]  eta: 0:04:46  lr: 0.000108  grad: 0.0837 (0.0875)  loss: 0.8158 (0.8147)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [27]  [5300/6250]  eta: 0:04:19  lr: 0.000108  grad: 0.0844 (0.0875)  loss: 0.8115 (0.8147)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [27]  [5400/6250]  eta: 0:03:51  lr: 0.000108  grad: 0.0864 (0.0875)  loss: 0.8129 (0.8147)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [27]  [5500/6250]  eta: 0:03:24  lr: 0.000108  grad: 0.0900 (0.0875)  loss: 0.8143 (0.8147)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [27]  [5600/6250]  eta: 0:02:57  lr: 0.000108  grad: 0.0829 (0.0876)  loss: 0.8144 (0.8147)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [27]  [5700/6250]  eta: 0:02:29  lr: 0.000108  grad: 0.0851 (0.0876)  loss: 0.8153 (0.8147)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [27]  [5800/6250]  eta: 0:02:02  lr: 0.000108  grad: 0.0909 (0.0876)  loss: 0.8111 (0.8147)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [27]  [5900/6250]  eta: 0:01:35  lr: 0.000108  grad: 0.0821 (0.0876)  loss: 0.8126 (0.8147)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [27]  [6000/6250]  eta: 0:01:08  lr: 0.000108  grad: 0.0802 (0.0876)  loss: 0.8122 (0.8147)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [27]  [6100/6250]  eta: 0:00:40  lr: 0.000108  grad: 0.0818 (0.0875)  loss: 0.8185 (0.8148)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [27]  [6200/6250]  eta: 0:00:13  lr: 0.000108  grad: 0.0876 (0.0875)  loss: 0.8223 (0.8148)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [27]  [6249/6250]  eta: 0:00:00  lr: 0.000108  grad: 0.0865 (0.0875)  loss: 0.8258 (0.8148)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [27] Total time: 0:28:27 (0.2732 s / it)
Averaged stats: lr: 0.000108  grad: 0.0865 (0.0875)  loss: 0.8258 (0.8148)
Eval (hcp-train-subset): [27]  [ 0/62]  eta: 0:03:55  loss: 0.8464 (0.8464)  time: 3.7922  data: 3.6473  max mem: 26157
Eval (hcp-train-subset): [27]  [61/62]  eta: 0:00:00  loss: 0.8320 (0.8338)  time: 0.1171  data: 0.0340  max mem: 26157
Eval (hcp-train-subset): [27] Total time: 0:00:12 (0.2026 s / it)
Averaged stats (hcp-train-subset): loss: 0.8320 (0.8338)
Making plots (hcp-train-subset): example=3
Eval (hcp-val): [27]  [ 0/62]  eta: 0:05:43  loss: 0.8243 (0.8243)  time: 5.5353  data: 5.4511  max mem: 26157
Eval (hcp-val): [27]  [61/62]  eta: 0:00:00  loss: 0.8297 (0.8300)  time: 0.1071  data: 0.0224  max mem: 26157
Eval (hcp-val): [27] Total time: 0:00:12 (0.2018 s / it)
Averaged stats (hcp-val): loss: 0.8297 (0.8300)
Making plots (hcp-val): example=51
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [28]  [   0/6250]  eta: 10:18:35  lr: 0.000108  grad: 0.0687 (0.0687)  loss: 0.8449 (0.8449)  time: 5.9385  data: 5.6620  max mem: 26157
Train: [28]  [ 100/6250]  eta: 0:33:28  lr: 0.000108  grad: 0.0785 (0.0932)  loss: 0.8257 (0.8283)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [28]  [ 200/6250]  eta: 0:30:12  lr: 0.000108  grad: 0.0873 (0.0910)  loss: 0.8199 (0.8248)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [28]  [ 300/6250]  eta: 0:28:47  lr: 0.000108  grad: 0.0856 (0.0907)  loss: 0.8191 (0.8225)  time: 0.2701  data: 0.0003  max mem: 26157
Train: [28]  [ 400/6250]  eta: 0:28:43  lr: 0.000108  grad: 0.0842 (0.0901)  loss: 0.8145 (0.8206)  time: 0.3404  data: 0.0625  max mem: 26157
Train: [28]  [ 500/6250]  eta: 0:27:49  lr: 0.000108  grad: 0.0779 (0.0884)  loss: 0.8292 (0.8207)  time: 0.2702  data: 0.0003  max mem: 26157
Train: [28]  [ 600/6250]  eta: 0:27:02  lr: 0.000108  grad: 0.0826 (0.0875)  loss: 0.8180 (0.8207)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [28]  [ 700/6250]  eta: 0:26:22  lr: 0.000108  grad: 0.0813 (0.0868)  loss: 0.8187 (0.8205)  time: 0.2743  data: 0.0002  max mem: 26157
Train: [28]  [ 800/6250]  eta: 0:26:05  lr: 0.000108  grad: 0.0773 (0.0863)  loss: 0.8215 (0.8202)  time: 0.3269  data: 0.0528  max mem: 26157
Train: [28]  [ 900/6250]  eta: 0:25:26  lr: 0.000108  grad: 0.0842 (0.0857)  loss: 0.8197 (0.8201)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [28]  [1000/6250]  eta: 0:24:51  lr: 0.000108  grad: 0.0835 (0.0857)  loss: 0.8121 (0.8199)  time: 0.2701  data: 0.0003  max mem: 26157
Train: [28]  [1100/6250]  eta: 0:24:16  lr: 0.000108  grad: 0.0830 (0.0853)  loss: 0.8143 (0.8196)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [28]  [1200/6250]  eta: 0:23:43  lr: 0.000108  grad: 0.0793 (0.0854)  loss: 0.8123 (0.8192)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [28]  [1300/6250]  eta: 0:23:10  lr: 0.000107  grad: 0.0846 (0.0854)  loss: 0.8112 (0.8189)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [28]  [1400/6250]  eta: 0:22:40  lr: 0.000107  grad: 0.0799 (0.0852)  loss: 0.8182 (0.8187)  time: 0.2779  data: 0.0002  max mem: 26157
Train: [28]  [1500/6250]  eta: 0:22:08  lr: 0.000107  grad: 0.0843 (0.0853)  loss: 0.8203 (0.8185)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [28]  [1600/6250]  eta: 0:21:37  lr: 0.000107  grad: 0.0834 (0.0853)  loss: 0.8161 (0.8184)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [28]  [1700/6250]  eta: 0:21:07  lr: 0.000107  grad: 0.0877 (0.0855)  loss: 0.8158 (0.8181)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [28]  [1800/6250]  eta: 0:20:37  lr: 0.000107  grad: 0.0855 (0.0859)  loss: 0.8202 (0.8179)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [28]  [1900/6250]  eta: 0:20:19  lr: 0.000107  grad: 0.0810 (0.0858)  loss: 0.8215 (0.8178)  time: 0.5068  data: 0.2323  max mem: 26157
Train: [28]  [2000/6250]  eta: 0:19:49  lr: 0.000107  grad: 0.0857 (0.0857)  loss: 0.8167 (0.8178)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [28]  [2100/6250]  eta: 0:19:25  lr: 0.000107  grad: 0.0771 (0.0856)  loss: 0.8194 (0.8179)  time: 0.4142  data: 0.1334  max mem: 26157
Train: [28]  [2200/6250]  eta: 0:18:54  lr: 0.000107  grad: 0.0812 (0.0854)  loss: 0.8227 (0.8179)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [28]  [2300/6250]  eta: 0:18:31  lr: 0.000107  grad: 0.0860 (0.0854)  loss: 0.8210 (0.8178)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [28]  [2400/6250]  eta: 0:18:01  lr: 0.000107  grad: 0.0830 (0.0855)  loss: 0.8163 (0.8178)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [28]  [2500/6250]  eta: 0:17:31  lr: 0.000107  grad: 0.0805 (0.0855)  loss: 0.8164 (0.8177)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [28]  [2600/6250]  eta: 0:17:02  lr: 0.000107  grad: 0.0853 (0.0857)  loss: 0.8130 (0.8175)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [28]  [2700/6250]  eta: 0:16:32  lr: 0.000107  grad: 0.0863 (0.0858)  loss: 0.8133 (0.8173)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [28]  [2800/6250]  eta: 0:16:03  lr: 0.000107  grad: 0.0900 (0.0858)  loss: 0.8063 (0.8171)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [28]  [2900/6250]  eta: 0:15:34  lr: 0.000107  grad: 0.0835 (0.0859)  loss: 0.8186 (0.8169)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [28]  [3000/6250]  eta: 0:15:05  lr: 0.000107  grad: 0.0843 (0.0859)  loss: 0.8125 (0.8168)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [28]  [3100/6250]  eta: 0:14:36  lr: 0.000107  grad: 0.0832 (0.0861)  loss: 0.8125 (0.8167)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [28]  [3200/6250]  eta: 0:14:07  lr: 0.000107  grad: 0.0865 (0.0861)  loss: 0.8073 (0.8166)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [28]  [3300/6250]  eta: 0:13:39  lr: 0.000107  grad: 0.0929 (0.0862)  loss: 0.8092 (0.8165)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [28]  [3400/6250]  eta: 0:13:10  lr: 0.000107  grad: 0.0821 (0.0863)  loss: 0.8175 (0.8164)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [28]  [3500/6250]  eta: 0:12:42  lr: 0.000107  grad: 0.0846 (0.0863)  loss: 0.8147 (0.8164)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [28]  [3600/6250]  eta: 0:12:14  lr: 0.000107  grad: 0.0810 (0.0863)  loss: 0.8162 (0.8164)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [28]  [3700/6250]  eta: 0:11:46  lr: 0.000107  grad: 0.0826 (0.0863)  loss: 0.8101 (0.8163)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [28]  [3800/6250]  eta: 0:11:17  lr: 0.000107  grad: 0.0865 (0.0865)  loss: 0.8180 (0.8162)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [28]  [3900/6250]  eta: 0:10:49  lr: 0.000107  grad: 0.0874 (0.0865)  loss: 0.8128 (0.8161)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [28]  [4000/6250]  eta: 0:10:21  lr: 0.000107  grad: 0.0868 (0.0866)  loss: 0.8116 (0.8161)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [28]  [4100/6250]  eta: 0:09:53  lr: 0.000107  grad: 0.0876 (0.0867)  loss: 0.8161 (0.8160)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [28]  [4200/6250]  eta: 0:09:25  lr: 0.000107  grad: 0.0807 (0.0867)  loss: 0.8174 (0.8161)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [28]  [4300/6250]  eta: 0:08:59  lr: 0.000107  grad: 0.0811 (0.0867)  loss: 0.8148 (0.8161)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [28]  [4400/6250]  eta: 0:08:32  lr: 0.000107  grad: 0.0848 (0.0867)  loss: 0.8200 (0.8161)  time: 0.3681  data: 0.0828  max mem: 26157
Train: [28]  [4500/6250]  eta: 0:08:04  lr: 0.000107  grad: 0.0875 (0.0867)  loss: 0.8145 (0.8161)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [28]  [4600/6250]  eta: 0:07:36  lr: 0.000107  grad: 0.0863 (0.0867)  loss: 0.8150 (0.8161)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [28]  [4700/6250]  eta: 0:07:08  lr: 0.000107  grad: 0.0857 (0.0867)  loss: 0.8123 (0.8161)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [28]  [4800/6250]  eta: 0:06:40  lr: 0.000107  grad: 0.0816 (0.0867)  loss: 0.8150 (0.8161)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [28]  [4900/6250]  eta: 0:06:13  lr: 0.000107  grad: 0.0838 (0.0867)  loss: 0.8204 (0.8162)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [28]  [5000/6250]  eta: 0:05:46  lr: 0.000107  grad: 0.0830 (0.0867)  loss: 0.8192 (0.8162)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [28]  [5100/6250]  eta: 0:05:18  lr: 0.000107  grad: 0.0830 (0.0867)  loss: 0.8145 (0.8162)  time: 0.3457  data: 0.0714  max mem: 26157
Train: [28]  [5200/6250]  eta: 0:04:51  lr: 0.000107  grad: 0.0851 (0.0868)  loss: 0.8190 (0.8162)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [28]  [5300/6250]  eta: 0:04:23  lr: 0.000107  grad: 0.0841 (0.0868)  loss: 0.8181 (0.8163)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [28]  [5400/6250]  eta: 0:03:55  lr: 0.000107  grad: 0.0841 (0.0868)  loss: 0.8209 (0.8163)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [28]  [5500/6250]  eta: 0:03:27  lr: 0.000107  grad: 0.0852 (0.0869)  loss: 0.8130 (0.8163)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [28]  [5600/6250]  eta: 0:02:59  lr: 0.000106  grad: 0.0845 (0.0869)  loss: 0.8181 (0.8163)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [28]  [5700/6250]  eta: 0:02:32  lr: 0.000106  grad: 0.0889 (0.0870)  loss: 0.8198 (0.8164)  time: 0.2696  data: 0.0001  max mem: 26157
Train: [28]  [5800/6250]  eta: 0:02:04  lr: 0.000106  grad: 0.0819 (0.0870)  loss: 0.8174 (0.8164)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [28]  [5900/6250]  eta: 0:01:36  lr: 0.000106  grad: 0.0809 (0.0870)  loss: 0.8187 (0.8164)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [28]  [6000/6250]  eta: 0:01:09  lr: 0.000106  grad: 0.0897 (0.0870)  loss: 0.8155 (0.8164)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [28]  [6100/6250]  eta: 0:00:41  lr: 0.000106  grad: 0.0851 (0.0870)  loss: 0.8202 (0.8164)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [28]  [6200/6250]  eta: 0:00:13  lr: 0.000106  grad: 0.0789 (0.0869)  loss: 0.8155 (0.8164)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [28]  [6249/6250]  eta: 0:00:00  lr: 0.000106  grad: 0.0803 (0.0869)  loss: 0.8216 (0.8165)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [28] Total time: 0:28:58 (0.2782 s / it)
Averaged stats: lr: 0.000106  grad: 0.0803 (0.0869)  loss: 0.8216 (0.8165)
Eval (hcp-train-subset): [28]  [ 0/62]  eta: 0:03:35  loss: 0.8439 (0.8439)  time: 3.4767  data: 3.3645  max mem: 26157
Eval (hcp-train-subset): [28]  [61/62]  eta: 0:00:00  loss: 0.8319 (0.8351)  time: 0.1230  data: 0.0384  max mem: 26157
Eval (hcp-train-subset): [28] Total time: 0:00:12 (0.2092 s / it)
Averaged stats (hcp-train-subset): loss: 0.8319 (0.8351)
Making plots (hcp-train-subset): example=53
Eval (hcp-val): [28]  [ 0/62]  eta: 0:04:46  loss: 0.8274 (0.8274)  time: 4.6218  data: 4.5377  max mem: 26157
Eval (hcp-val): [28]  [61/62]  eta: 0:00:00  loss: 0.8294 (0.8299)  time: 0.1306  data: 0.0456  max mem: 26157
Eval (hcp-val): [28] Total time: 0:00:12 (0.2068 s / it)
Averaged stats (hcp-val): loss: 0.8294 (0.8299)
Making plots (hcp-val): example=47
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [29]  [   0/6250]  eta: 11:18:07  lr: 0.000106  grad: 0.0932 (0.0932)  loss: 0.8476 (0.8476)  time: 6.5099  data: 6.2315  max mem: 26157
Train: [29]  [ 100/6250]  eta: 0:33:59  lr: 0.000106  grad: 0.0827 (0.0997)  loss: 0.8244 (0.8290)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [29]  [ 200/6250]  eta: 0:30:18  lr: 0.000106  grad: 0.0831 (0.0935)  loss: 0.8089 (0.8237)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [29]  [ 300/6250]  eta: 0:28:46  lr: 0.000106  grad: 0.0846 (0.0916)  loss: 0.8154 (0.8210)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [29]  [ 400/6250]  eta: 0:27:51  lr: 0.000106  grad: 0.0831 (0.0912)  loss: 0.8164 (0.8202)  time: 0.2747  data: 0.0002  max mem: 26157
Train: [29]  [ 500/6250]  eta: 0:27:06  lr: 0.000106  grad: 0.0806 (0.0900)  loss: 0.8145 (0.8202)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [29]  [ 600/6250]  eta: 0:26:26  lr: 0.000106  grad: 0.0786 (0.0890)  loss: 0.8197 (0.8205)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [29]  [ 700/6250]  eta: 0:25:50  lr: 0.000106  grad: 0.0814 (0.0885)  loss: 0.8251 (0.8210)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [29]  [ 800/6250]  eta: 0:25:16  lr: 0.000106  grad: 0.0853 (0.0882)  loss: 0.8226 (0.8209)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [29]  [ 900/6250]  eta: 0:24:46  lr: 0.000106  grad: 0.0778 (0.0878)  loss: 0.8265 (0.8210)  time: 0.2796  data: 0.0002  max mem: 26157
Train: [29]  [1000/6250]  eta: 0:24:16  lr: 0.000106  grad: 0.0801 (0.0875)  loss: 0.8153 (0.8206)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [29]  [1100/6250]  eta: 0:23:44  lr: 0.000106  grad: 0.0851 (0.0875)  loss: 0.8097 (0.8203)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [29]  [1200/6250]  eta: 0:23:13  lr: 0.000106  grad: 0.0870 (0.0877)  loss: 0.8179 (0.8198)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [29]  [1300/6250]  eta: 0:22:43  lr: 0.000106  grad: 0.0835 (0.0875)  loss: 0.8149 (0.8195)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [29]  [1400/6250]  eta: 0:22:14  lr: 0.000106  grad: 0.0895 (0.0877)  loss: 0.8025 (0.8189)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [29]  [1500/6250]  eta: 0:21:44  lr: 0.000106  grad: 0.0893 (0.0880)  loss: 0.8173 (0.8185)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [29]  [1600/6250]  eta: 0:21:21  lr: 0.000106  grad: 0.0823 (0.0880)  loss: 0.8140 (0.8182)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [29]  [1700/6250]  eta: 0:20:57  lr: 0.000106  grad: 0.0844 (0.0881)  loss: 0.8171 (0.8179)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [29]  [1800/6250]  eta: 0:20:28  lr: 0.000106  grad: 0.0889 (0.0881)  loss: 0.8153 (0.8176)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [29]  [1900/6250]  eta: 0:19:59  lr: 0.000106  grad: 0.0821 (0.0881)  loss: 0.8079 (0.8173)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [29]  [2000/6250]  eta: 0:19:30  lr: 0.000106  grad: 0.0853 (0.0881)  loss: 0.8108 (0.8171)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [29]  [2100/6250]  eta: 0:19:02  lr: 0.000106  grad: 0.0837 (0.0882)  loss: 0.8114 (0.8169)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [29]  [2200/6250]  eta: 0:18:33  lr: 0.000106  grad: 0.0868 (0.0881)  loss: 0.8132 (0.8167)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [29]  [2300/6250]  eta: 0:18:05  lr: 0.000106  grad: 0.0874 (0.0881)  loss: 0.8157 (0.8166)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [29]  [2400/6250]  eta: 0:17:37  lr: 0.000106  grad: 0.0934 (0.0881)  loss: 0.8065 (0.8165)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [29]  [2500/6250]  eta: 0:17:09  lr: 0.000106  grad: 0.0831 (0.0881)  loss: 0.8167 (0.8163)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [29]  [2600/6250]  eta: 0:16:41  lr: 0.000106  grad: 0.0898 (0.0883)  loss: 0.8118 (0.8160)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [29]  [2700/6250]  eta: 0:16:14  lr: 0.000106  grad: 0.0827 (0.0883)  loss: 0.8107 (0.8158)  time: 0.3179  data: 0.0461  max mem: 26157
Train: [29]  [2800/6250]  eta: 0:15:46  lr: 0.000106  grad: 0.0909 (0.0885)  loss: 0.8126 (0.8156)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [29]  [2900/6250]  eta: 0:15:18  lr: 0.000106  grad: 0.0862 (0.0886)  loss: 0.8098 (0.8153)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [29]  [3000/6250]  eta: 0:14:50  lr: 0.000106  grad: 0.0882 (0.0886)  loss: 0.8135 (0.8151)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [29]  [3100/6250]  eta: 0:14:22  lr: 0.000106  grad: 0.0874 (0.0886)  loss: 0.8134 (0.8150)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [29]  [3200/6250]  eta: 0:13:54  lr: 0.000106  grad: 0.0854 (0.0886)  loss: 0.8087 (0.8149)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [29]  [3300/6250]  eta: 0:13:27  lr: 0.000106  grad: 0.0838 (0.0886)  loss: 0.8140 (0.8149)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [29]  [3400/6250]  eta: 0:12:59  lr: 0.000106  grad: 0.0835 (0.0886)  loss: 0.8066 (0.8148)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [29]  [3500/6250]  eta: 0:12:31  lr: 0.000105  grad: 0.0845 (0.0886)  loss: 0.8134 (0.8147)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [29]  [3600/6250]  eta: 0:12:04  lr: 0.000105  grad: 0.0850 (0.0886)  loss: 0.8141 (0.8147)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [29]  [3700/6250]  eta: 0:11:36  lr: 0.000105  grad: 0.0907 (0.0886)  loss: 0.8151 (0.8147)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [29]  [3800/6250]  eta: 0:11:09  lr: 0.000105  grad: 0.0912 (0.0888)  loss: 0.8160 (0.8147)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [29]  [3900/6250]  eta: 0:10:41  lr: 0.000105  grad: 0.0852 (0.0887)  loss: 0.8137 (0.8147)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [29]  [4000/6250]  eta: 0:10:14  lr: 0.000105  grad: 0.0920 (0.0889)  loss: 0.8154 (0.8147)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [29]  [4100/6250]  eta: 0:09:46  lr: 0.000105  grad: 0.0865 (0.0891)  loss: 0.8131 (0.8147)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [29]  [4200/6250]  eta: 0:09:19  lr: 0.000105  grad: 0.0804 (0.0890)  loss: 0.8180 (0.8147)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [29]  [4300/6250]  eta: 0:08:51  lr: 0.000105  grad: 0.0873 (0.0891)  loss: 0.8080 (0.8147)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [29]  [4400/6250]  eta: 0:08:24  lr: 0.000105  grad: 0.0805 (0.0891)  loss: 0.8132 (0.8146)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [29]  [4500/6250]  eta: 0:07:56  lr: 0.000105  grad: 0.0900 (0.0891)  loss: 0.8001 (0.8146)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [29]  [4600/6250]  eta: 0:07:29  lr: 0.000105  grad: 0.0909 (0.0891)  loss: 0.8181 (0.8145)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [29]  [4700/6250]  eta: 0:07:02  lr: 0.000105  grad: 0.0841 (0.0891)  loss: 0.8142 (0.8145)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [29]  [4800/6250]  eta: 0:06:34  lr: 0.000105  grad: 0.0836 (0.0891)  loss: 0.8168 (0.8145)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [29]  [4900/6250]  eta: 0:06:07  lr: 0.000105  grad: 0.0889 (0.0891)  loss: 0.8135 (0.8145)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [29]  [5000/6250]  eta: 0:05:40  lr: 0.000105  grad: 0.0864 (0.0891)  loss: 0.8142 (0.8145)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [29]  [5100/6250]  eta: 0:05:12  lr: 0.000105  grad: 0.0817 (0.0890)  loss: 0.8141 (0.8145)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [29]  [5200/6250]  eta: 0:04:45  lr: 0.000105  grad: 0.0871 (0.0890)  loss: 0.8144 (0.8145)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [29]  [5300/6250]  eta: 0:04:18  lr: 0.000105  grad: 0.0820 (0.0890)  loss: 0.8124 (0.8145)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [29]  [5400/6250]  eta: 0:03:51  lr: 0.000105  grad: 0.0851 (0.0890)  loss: 0.8168 (0.8146)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [29]  [5500/6250]  eta: 0:03:23  lr: 0.000105  grad: 0.0821 (0.0890)  loss: 0.8194 (0.8147)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [29]  [5600/6250]  eta: 0:02:57  lr: 0.000105  grad: 0.0851 (0.0889)  loss: 0.8146 (0.8147)  time: 0.3857  data: 0.1082  max mem: 26157
Train: [29]  [5700/6250]  eta: 0:02:30  lr: 0.000105  grad: 0.0893 (0.0890)  loss: 0.8150 (0.8147)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [29]  [5800/6250]  eta: 0:02:02  lr: 0.000105  grad: 0.0851 (0.0892)  loss: 0.8216 (0.8147)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [29]  [5900/6250]  eta: 0:01:35  lr: 0.000105  grad: 0.0857 (0.0892)  loss: 0.8142 (0.8147)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [29]  [6000/6250]  eta: 0:01:08  lr: 0.000105  grad: 0.0862 (0.0892)  loss: 0.8159 (0.8147)  time: 0.2720  data: 0.0001  max mem: 26157
Train: [29]  [6100/6250]  eta: 0:00:40  lr: 0.000105  grad: 0.0911 (0.0894)  loss: 0.8159 (0.8147)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [29]  [6200/6250]  eta: 0:00:13  lr: 0.000105  grad: 0.0889 (0.0894)  loss: 0.8167 (0.8147)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [29]  [6249/6250]  eta: 0:00:00  lr: 0.000105  grad: 0.0858 (0.0894)  loss: 0.8129 (0.8147)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [29] Total time: 0:28:31 (0.2739 s / it)
Averaged stats: lr: 0.000105  grad: 0.0858 (0.0894)  loss: 0.8129 (0.8147)
Eval (hcp-train-subset): [29]  [ 0/62]  eta: 0:03:48  loss: 0.8464 (0.8464)  time: 3.6790  data: 3.5529  max mem: 26157
Eval (hcp-train-subset): [29]  [61/62]  eta: 0:00:00  loss: 0.8328 (0.8341)  time: 0.1362  data: 0.0533  max mem: 26157
Eval (hcp-train-subset): [29] Total time: 0:00:13 (0.2108 s / it)
Averaged stats (hcp-train-subset): loss: 0.8328 (0.8341)
Making plots (hcp-train-subset): example=15
Eval (hcp-val): [29]  [ 0/62]  eta: 0:04:54  loss: 0.8253 (0.8253)  time: 4.7499  data: 4.6662  max mem: 26157
Eval (hcp-val): [29]  [61/62]  eta: 0:00:00  loss: 0.8290 (0.8294)  time: 0.1283  data: 0.0449  max mem: 26157
Eval (hcp-val): [29] Total time: 0:00:12 (0.2066 s / it)
Averaged stats (hcp-val): loss: 0.8290 (0.8294)
Making plots (hcp-val): example=8
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [30]  [   0/6250]  eta: 7:26:34  lr: 0.000105  grad: 0.0570 (0.0570)  loss: 0.8668 (0.8668)  time: 4.2871  data: 3.9217  max mem: 26157
Train: [30]  [ 100/6250]  eta: 0:34:07  lr: 0.000105  grad: 0.1053 (0.1077)  loss: 0.8053 (0.8261)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [30]  [ 200/6250]  eta: 0:30:23  lr: 0.000105  grad: 0.0908 (0.1014)  loss: 0.8223 (0.8225)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [30]  [ 300/6250]  eta: 0:28:51  lr: 0.000105  grad: 0.0885 (0.1001)  loss: 0.8122 (0.8203)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [30]  [ 400/6250]  eta: 0:27:52  lr: 0.000105  grad: 0.0822 (0.0978)  loss: 0.8182 (0.8193)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [30]  [ 500/6250]  eta: 0:27:05  lr: 0.000105  grad: 0.0904 (0.0965)  loss: 0.8189 (0.8182)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [30]  [ 600/6250]  eta: 0:26:25  lr: 0.000105  grad: 0.0808 (0.0956)  loss: 0.8187 (0.8175)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [30]  [ 700/6250]  eta: 0:25:49  lr: 0.000105  grad: 0.0814 (0.0947)  loss: 0.8206 (0.8170)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [30]  [ 800/6250]  eta: 0:25:15  lr: 0.000105  grad: 0.0808 (0.0938)  loss: 0.8182 (0.8168)  time: 0.2692  data: 0.0001  max mem: 26157
Train: [30]  [ 900/6250]  eta: 0:24:42  lr: 0.000105  grad: 0.0828 (0.0931)  loss: 0.8147 (0.8166)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [30]  [1000/6250]  eta: 0:24:10  lr: 0.000105  grad: 0.0842 (0.0926)  loss: 0.8216 (0.8165)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [30]  [1100/6250]  eta: 0:23:39  lr: 0.000105  grad: 0.0823 (0.0920)  loss: 0.8133 (0.8163)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [30]  [1200/6250]  eta: 0:23:09  lr: 0.000105  grad: 0.0828 (0.0918)  loss: 0.8133 (0.8161)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [30]  [1300/6250]  eta: 0:22:39  lr: 0.000105  grad: 0.0888 (0.0927)  loss: 0.8138 (0.8158)  time: 0.2697  data: 0.0001  max mem: 26157
Train: [30]  [1400/6250]  eta: 0:22:10  lr: 0.000104  grad: 0.0829 (0.0923)  loss: 0.8119 (0.8156)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [30]  [1500/6250]  eta: 0:21:41  lr: 0.000104  grad: 0.0860 (0.0921)  loss: 0.8211 (0.8157)  time: 0.2689  data: 0.0004  max mem: 26157
Train: [30]  [1600/6250]  eta: 0:21:12  lr: 0.000104  grad: 0.0898 (0.0918)  loss: 0.8135 (0.8156)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [30]  [1700/6250]  eta: 0:20:44  lr: 0.000104  grad: 0.0851 (0.0915)  loss: 0.8129 (0.8157)  time: 0.2764  data: 0.0002  max mem: 26157
Train: [30]  [1800/6250]  eta: 0:20:29  lr: 0.000104  grad: 0.0859 (0.0915)  loss: 0.8161 (0.8156)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [30]  [1900/6250]  eta: 0:20:00  lr: 0.000104  grad: 0.0793 (0.0912)  loss: 0.8220 (0.8156)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [30]  [2000/6250]  eta: 0:19:40  lr: 0.000104  grad: 0.0881 (0.0909)  loss: 0.8114 (0.8157)  time: 0.4830  data: 0.2094  max mem: 26157
Train: [30]  [2100/6250]  eta: 0:19:21  lr: 0.000104  grad: 0.0880 (0.0908)  loss: 0.8140 (0.8156)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [30]  [2200/6250]  eta: 0:18:56  lr: 0.000104  grad: 0.0873 (0.0906)  loss: 0.8139 (0.8156)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [30]  [2300/6250]  eta: 0:18:26  lr: 0.000104  grad: 0.0837 (0.0905)  loss: 0.8135 (0.8156)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [30]  [2400/6250]  eta: 0:17:57  lr: 0.000104  grad: 0.0857 (0.0903)  loss: 0.8159 (0.8156)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [30]  [2500/6250]  eta: 0:17:27  lr: 0.000104  grad: 0.0859 (0.0902)  loss: 0.8178 (0.8155)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [30]  [2600/6250]  eta: 0:16:58  lr: 0.000104  grad: 0.0849 (0.0901)  loss: 0.8165 (0.8156)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [30]  [2700/6250]  eta: 0:16:29  lr: 0.000104  grad: 0.0818 (0.0900)  loss: 0.8187 (0.8157)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [30]  [2800/6250]  eta: 0:16:09  lr: 0.000104  grad: 0.0825 (0.0899)  loss: 0.8272 (0.8159)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [30]  [2900/6250]  eta: 0:15:40  lr: 0.000104  grad: 0.0848 (0.0899)  loss: 0.8188 (0.8159)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [30]  [3000/6250]  eta: 0:15:10  lr: 0.000104  grad: 0.0864 (0.0898)  loss: 0.8211 (0.8159)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [30]  [3100/6250]  eta: 0:14:41  lr: 0.000104  grad: 0.0872 (0.0898)  loss: 0.8107 (0.8159)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [30]  [3200/6250]  eta: 0:14:12  lr: 0.000104  grad: 0.0900 (0.0897)  loss: 0.8161 (0.8159)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [30]  [3300/6250]  eta: 0:13:44  lr: 0.000104  grad: 0.0829 (0.0897)  loss: 0.8155 (0.8159)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [30]  [3400/6250]  eta: 0:13:15  lr: 0.000104  grad: 0.0856 (0.0896)  loss: 0.8152 (0.8160)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [30]  [3500/6250]  eta: 0:12:46  lr: 0.000104  grad: 0.0885 (0.0896)  loss: 0.8145 (0.8160)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [30]  [3600/6250]  eta: 0:12:22  lr: 0.000104  grad: 0.0828 (0.0897)  loss: 0.8196 (0.8160)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [30]  [3700/6250]  eta: 0:11:59  lr: 0.000104  grad: 0.0844 (0.0897)  loss: 0.8129 (0.8160)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [30]  [3800/6250]  eta: 0:11:30  lr: 0.000104  grad: 0.0875 (0.0896)  loss: 0.8138 (0.8160)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [30]  [3900/6250]  eta: 0:11:01  lr: 0.000104  grad: 0.0925 (0.0896)  loss: 0.8180 (0.8160)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [30]  [4000/6250]  eta: 0:10:33  lr: 0.000104  grad: 0.0836 (0.0895)  loss: 0.8186 (0.8160)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [30]  [4100/6250]  eta: 0:10:04  lr: 0.000104  grad: 0.0856 (0.0895)  loss: 0.8148 (0.8160)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [30]  [4200/6250]  eta: 0:09:35  lr: 0.000104  grad: 0.0880 (0.0895)  loss: 0.8215 (0.8160)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [30]  [4300/6250]  eta: 0:09:06  lr: 0.000104  grad: 0.0912 (0.0896)  loss: 0.8134 (0.8160)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [30]  [4400/6250]  eta: 0:08:38  lr: 0.000104  grad: 0.0898 (0.0897)  loss: 0.8173 (0.8160)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [30]  [4500/6250]  eta: 0:08:09  lr: 0.000104  grad: 0.0877 (0.0898)  loss: 0.8079 (0.8160)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [30]  [4600/6250]  eta: 0:07:41  lr: 0.000104  grad: 0.0876 (0.0898)  loss: 0.8118 (0.8160)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [30]  [4700/6250]  eta: 0:07:13  lr: 0.000104  grad: 0.0836 (0.0899)  loss: 0.8191 (0.8160)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [30]  [4800/6250]  eta: 0:06:44  lr: 0.000104  grad: 0.0920 (0.0899)  loss: 0.8113 (0.8159)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [30]  [4900/6250]  eta: 0:06:16  lr: 0.000104  grad: 0.0893 (0.0901)  loss: 0.8172 (0.8158)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [30]  [5000/6250]  eta: 0:05:48  lr: 0.000104  grad: 0.0869 (0.0902)  loss: 0.8186 (0.8158)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [30]  [5100/6250]  eta: 0:05:20  lr: 0.000104  grad: 0.0840 (0.0902)  loss: 0.8160 (0.8157)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [30]  [5200/6250]  eta: 0:04:52  lr: 0.000104  grad: 0.0847 (0.0902)  loss: 0.8152 (0.8157)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [30]  [5300/6250]  eta: 0:04:24  lr: 0.000104  grad: 0.0846 (0.0902)  loss: 0.8123 (0.8157)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [30]  [5400/6250]  eta: 0:03:56  lr: 0.000103  grad: 0.0898 (0.0902)  loss: 0.8168 (0.8157)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [30]  [5500/6250]  eta: 0:03:28  lr: 0.000103  grad: 0.0877 (0.0903)  loss: 0.8143 (0.8156)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [30]  [5600/6250]  eta: 0:03:00  lr: 0.000103  grad: 0.0873 (0.0903)  loss: 0.8179 (0.8156)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [30]  [5700/6250]  eta: 0:02:32  lr: 0.000103  grad: 0.0877 (0.0903)  loss: 0.8146 (0.8156)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [30]  [5800/6250]  eta: 0:02:04  lr: 0.000103  grad: 0.0974 (0.0903)  loss: 0.8092 (0.8156)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [30]  [5900/6250]  eta: 0:01:37  lr: 0.000103  grad: 0.0873 (0.0904)  loss: 0.8120 (0.8155)  time: 0.2776  data: 0.0003  max mem: 26157
Train: [30]  [6000/6250]  eta: 0:01:09  lr: 0.000103  grad: 0.0857 (0.0903)  loss: 0.8115 (0.8155)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [30]  [6100/6250]  eta: 0:00:41  lr: 0.000103  grad: 0.0917 (0.0903)  loss: 0.8105 (0.8155)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [30]  [6200/6250]  eta: 0:00:13  lr: 0.000103  grad: 0.0947 (0.0904)  loss: 0.8121 (0.8154)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [30]  [6249/6250]  eta: 0:00:00  lr: 0.000103  grad: 0.0883 (0.0904)  loss: 0.8114 (0.8153)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [30] Total time: 0:29:04 (0.2791 s / it)
Averaged stats: lr: 0.000103  grad: 0.0883 (0.0904)  loss: 0.8114 (0.8153)
Eval (hcp-train-subset): [30]  [ 0/62]  eta: 0:04:10  loss: 0.8474 (0.8474)  time: 4.0343  data: 3.9157  max mem: 26157
Eval (hcp-train-subset): [30]  [61/62]  eta: 0:00:00  loss: 0.8324 (0.8329)  time: 0.1165  data: 0.0315  max mem: 26157
Eval (hcp-train-subset): [30] Total time: 0:00:12 (0.2017 s / it)
Averaged stats (hcp-train-subset): loss: 0.8324 (0.8329)
Making plots (hcp-train-subset): example=43
Eval (hcp-val): [30]  [ 0/62]  eta: 0:04:41  loss: 0.8258 (0.8258)  time: 4.5404  data: 4.4563  max mem: 26157
Eval (hcp-val): [30]  [61/62]  eta: 0:00:00  loss: 0.8286 (0.8298)  time: 0.1311  data: 0.0478  max mem: 26157
Eval (hcp-val): [30] Total time: 0:00:13 (0.2124 s / it)
Averaged stats (hcp-val): loss: 0.8286 (0.8298)
Making plots (hcp-val): example=6
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [31]  [   0/6250]  eta: 10:14:42  lr: 0.000103  grad: 0.0812 (0.0812)  loss: 0.8323 (0.8323)  time: 5.9011  data: 5.6057  max mem: 26157
Train: [31]  [ 100/6250]  eta: 0:33:32  lr: 0.000103  grad: 0.0873 (0.1033)  loss: 0.8179 (0.8284)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [31]  [ 200/6250]  eta: 0:30:11  lr: 0.000103  grad: 0.0862 (0.0968)  loss: 0.8200 (0.8258)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [31]  [ 300/6250]  eta: 0:28:44  lr: 0.000103  grad: 0.0759 (0.0932)  loss: 0.8250 (0.8239)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [31]  [ 400/6250]  eta: 0:27:45  lr: 0.000103  grad: 0.0928 (0.0930)  loss: 0.8143 (0.8216)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [31]  [ 500/6250]  eta: 0:26:59  lr: 0.000103  grad: 0.0804 (0.0917)  loss: 0.8156 (0.8201)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [31]  [ 600/6250]  eta: 0:26:19  lr: 0.000103  grad: 0.0887 (0.0918)  loss: 0.8156 (0.8192)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [31]  [ 700/6250]  eta: 0:25:43  lr: 0.000103  grad: 0.0834 (0.0913)  loss: 0.8140 (0.8185)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [31]  [ 800/6250]  eta: 0:25:09  lr: 0.000103  grad: 0.0809 (0.0906)  loss: 0.8182 (0.8182)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [31]  [ 900/6250]  eta: 0:24:38  lr: 0.000103  grad: 0.0789 (0.0900)  loss: 0.8218 (0.8182)  time: 0.2753  data: 0.0002  max mem: 26157
Train: [31]  [1000/6250]  eta: 0:24:12  lr: 0.000103  grad: 0.0835 (0.0896)  loss: 0.8162 (0.8181)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [31]  [1100/6250]  eta: 0:23:41  lr: 0.000103  grad: 0.0828 (0.0891)  loss: 0.8180 (0.8181)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [31]  [1200/6250]  eta: 0:23:11  lr: 0.000103  grad: 0.0877 (0.0889)  loss: 0.8181 (0.8180)  time: 0.2751  data: 0.0002  max mem: 26157
Train: [31]  [1300/6250]  eta: 0:22:41  lr: 0.000103  grad: 0.0805 (0.0887)  loss: 0.8144 (0.8177)  time: 0.2695  data: 0.0003  max mem: 26157
Train: [31]  [1400/6250]  eta: 0:22:12  lr: 0.000103  grad: 0.0856 (0.0888)  loss: 0.8104 (0.8174)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [31]  [1500/6250]  eta: 0:21:42  lr: 0.000103  grad: 0.0887 (0.0887)  loss: 0.8118 (0.8170)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [31]  [1600/6250]  eta: 0:21:13  lr: 0.000103  grad: 0.0917 (0.0886)  loss: 0.8106 (0.8167)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [31]  [1700/6250]  eta: 0:20:45  lr: 0.000103  grad: 0.0846 (0.0888)  loss: 0.8134 (0.8165)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [31]  [1800/6250]  eta: 0:20:16  lr: 0.000103  grad: 0.0812 (0.0887)  loss: 0.8133 (0.8165)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [31]  [1900/6250]  eta: 0:19:48  lr: 0.000103  grad: 0.0889 (0.0886)  loss: 0.8144 (0.8164)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [31]  [2000/6250]  eta: 0:19:20  lr: 0.000103  grad: 0.0839 (0.0885)  loss: 0.8207 (0.8163)  time: 0.2693  data: 0.0001  max mem: 26157
Train: [31]  [2100/6250]  eta: 0:18:52  lr: 0.000103  grad: 0.0829 (0.0885)  loss: 0.8189 (0.8163)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [31]  [2200/6250]  eta: 0:18:24  lr: 0.000103  grad: 0.0862 (0.0884)  loss: 0.8158 (0.8163)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [31]  [2300/6250]  eta: 0:17:56  lr: 0.000103  grad: 0.0872 (0.0884)  loss: 0.8215 (0.8163)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [31]  [2400/6250]  eta: 0:17:29  lr: 0.000103  grad: 0.0840 (0.0883)  loss: 0.8115 (0.8163)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [31]  [2500/6250]  eta: 0:17:01  lr: 0.000103  grad: 0.0825 (0.0884)  loss: 0.8156 (0.8163)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [31]  [2600/6250]  eta: 0:16:33  lr: 0.000103  grad: 0.0927 (0.0885)  loss: 0.8133 (0.8162)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [31]  [2700/6250]  eta: 0:16:06  lr: 0.000103  grad: 0.0857 (0.0886)  loss: 0.8190 (0.8162)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [31]  [2800/6250]  eta: 0:15:38  lr: 0.000103  grad: 0.0839 (0.0886)  loss: 0.8221 (0.8162)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [31]  [2900/6250]  eta: 0:15:11  lr: 0.000103  grad: 0.0826 (0.0886)  loss: 0.8190 (0.8162)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [31]  [3000/6250]  eta: 0:14:43  lr: 0.000103  grad: 0.0943 (0.0887)  loss: 0.8179 (0.8162)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [31]  [3100/6250]  eta: 0:14:16  lr: 0.000103  grad: 0.0865 (0.0888)  loss: 0.8176 (0.8161)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [31]  [3200/6250]  eta: 0:13:49  lr: 0.000102  grad: 0.0831 (0.0888)  loss: 0.8165 (0.8161)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [31]  [3300/6250]  eta: 0:13:22  lr: 0.000102  grad: 0.0878 (0.0889)  loss: 0.8145 (0.8160)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [31]  [3400/6250]  eta: 0:12:54  lr: 0.000102  grad: 0.0839 (0.0890)  loss: 0.8171 (0.8160)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [31]  [3500/6250]  eta: 0:12:28  lr: 0.000102  grad: 0.0878 (0.0891)  loss: 0.8199 (0.8161)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [31]  [3600/6250]  eta: 0:12:01  lr: 0.000102  grad: 0.0920 (0.0891)  loss: 0.8115 (0.8160)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [31]  [3700/6250]  eta: 0:11:34  lr: 0.000102  grad: 0.0889 (0.0893)  loss: 0.8086 (0.8159)  time: 0.2741  data: 0.0002  max mem: 26157
Train: [31]  [3800/6250]  eta: 0:11:07  lr: 0.000102  grad: 0.0887 (0.0894)  loss: 0.8146 (0.8158)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [31]  [3900/6250]  eta: 0:10:39  lr: 0.000102  grad: 0.0987 (0.0896)  loss: 0.8024 (0.8156)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [31]  [4000/6250]  eta: 0:10:12  lr: 0.000102  grad: 0.0901 (0.0897)  loss: 0.8109 (0.8155)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [31]  [4100/6250]  eta: 0:09:44  lr: 0.000102  grad: 0.0903 (0.0898)  loss: 0.8119 (0.8154)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [31]  [4200/6250]  eta: 0:09:17  lr: 0.000102  grad: 0.0891 (0.0898)  loss: 0.8093 (0.8154)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [31]  [4300/6250]  eta: 0:08:50  lr: 0.000102  grad: 0.0963 (0.0899)  loss: 0.8132 (0.8153)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [31]  [4400/6250]  eta: 0:08:22  lr: 0.000102  grad: 0.0874 (0.0899)  loss: 0.8095 (0.8152)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [31]  [4500/6250]  eta: 0:07:55  lr: 0.000102  grad: 0.0886 (0.0899)  loss: 0.8100 (0.8151)  time: 0.2721  data: 0.0004  max mem: 26157
Train: [31]  [4600/6250]  eta: 0:07:28  lr: 0.000102  grad: 0.0936 (0.0900)  loss: 0.8130 (0.8150)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [31]  [4700/6250]  eta: 0:07:01  lr: 0.000102  grad: 0.0907 (0.0901)  loss: 0.8159 (0.8149)  time: 0.2779  data: 0.0002  max mem: 26157
Train: [31]  [4800/6250]  eta: 0:06:34  lr: 0.000102  grad: 0.0902 (0.0901)  loss: 0.8123 (0.8149)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [31]  [4900/6250]  eta: 0:06:06  lr: 0.000102  grad: 0.0871 (0.0901)  loss: 0.8112 (0.8148)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [31]  [5000/6250]  eta: 0:05:39  lr: 0.000102  grad: 0.0865 (0.0901)  loss: 0.8101 (0.8147)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [31]  [5100/6250]  eta: 0:05:12  lr: 0.000102  grad: 0.0868 (0.0901)  loss: 0.8111 (0.8147)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [31]  [5200/6250]  eta: 0:04:45  lr: 0.000102  grad: 0.0895 (0.0901)  loss: 0.8134 (0.8147)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [31]  [5300/6250]  eta: 0:04:17  lr: 0.000102  grad: 0.0912 (0.0902)  loss: 0.8128 (0.8147)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [31]  [5400/6250]  eta: 0:03:50  lr: 0.000102  grad: 0.0936 (0.0902)  loss: 0.8148 (0.8146)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [31]  [5500/6250]  eta: 0:03:23  lr: 0.000102  grad: 0.0924 (0.0902)  loss: 0.8108 (0.8146)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [31]  [5600/6250]  eta: 0:02:56  lr: 0.000102  grad: 0.0907 (0.0903)  loss: 0.8067 (0.8145)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [31]  [5700/6250]  eta: 0:02:29  lr: 0.000102  grad: 0.0917 (0.0903)  loss: 0.8092 (0.8145)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [31]  [5800/6250]  eta: 0:02:02  lr: 0.000102  grad: 0.0857 (0.0903)  loss: 0.8105 (0.8144)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [31]  [5900/6250]  eta: 0:01:35  lr: 0.000102  grad: 0.0923 (0.0904)  loss: 0.8101 (0.8143)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [31]  [6000/6250]  eta: 0:01:07  lr: 0.000102  grad: 0.0878 (0.0903)  loss: 0.8100 (0.8144)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [31]  [6100/6250]  eta: 0:00:40  lr: 0.000102  grad: 0.0862 (0.0903)  loss: 0.8096 (0.8143)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [31]  [6200/6250]  eta: 0:00:13  lr: 0.000102  grad: 0.0873 (0.0903)  loss: 0.8108 (0.8143)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [31]  [6249/6250]  eta: 0:00:00  lr: 0.000102  grad: 0.0847 (0.0903)  loss: 0.8193 (0.8143)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [31] Total time: 0:28:25 (0.2729 s / it)
Averaged stats: lr: 0.000102  grad: 0.0847 (0.0903)  loss: 0.8193 (0.8143)
Eval (hcp-train-subset): [31]  [ 0/62]  eta: 0:05:13  loss: 0.8341 (0.8341)  time: 5.0551  data: 4.9714  max mem: 26157
Eval (hcp-train-subset): [31]  [61/62]  eta: 0:00:00  loss: 0.8277 (0.8310)  time: 0.1284  data: 0.0457  max mem: 26157
Eval (hcp-train-subset): [31] Total time: 0:00:12 (0.2091 s / it)
Averaged stats (hcp-train-subset): loss: 0.8277 (0.8310)
Making plots (hcp-train-subset): example=2
Eval (hcp-val): [31]  [ 0/62]  eta: 0:04:15  loss: 0.8236 (0.8236)  time: 4.1175  data: 4.0128  max mem: 26157
Eval (hcp-val): [31]  [61/62]  eta: 0:00:00  loss: 0.8279 (0.8295)  time: 0.1425  data: 0.0598  max mem: 26157
Eval (hcp-val): [31] Total time: 0:00:13 (0.2207 s / it)
Averaged stats (hcp-val): loss: 0.8279 (0.8295)
Making plots (hcp-val): example=41
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [32]  [   0/6250]  eta: 7:57:20  lr: 0.000102  grad: 0.0622 (0.0622)  loss: 0.8541 (0.8541)  time: 4.5825  data: 4.2404  max mem: 26157
Train: [32]  [ 100/6250]  eta: 0:33:38  lr: 0.000102  grad: 0.0943 (0.0992)  loss: 0.8256 (0.8337)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [32]  [ 200/6250]  eta: 0:30:11  lr: 0.000102  grad: 0.0785 (0.1011)  loss: 0.8267 (0.8257)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [32]  [ 300/6250]  eta: 0:28:43  lr: 0.000102  grad: 0.0825 (0.0967)  loss: 0.8219 (0.8243)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [32]  [ 400/6250]  eta: 0:27:47  lr: 0.000102  grad: 0.0839 (0.0947)  loss: 0.8176 (0.8224)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [32]  [ 500/6250]  eta: 0:27:02  lr: 0.000102  grad: 0.0789 (0.0935)  loss: 0.8165 (0.8212)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [32]  [ 600/6250]  eta: 0:26:25  lr: 0.000102  grad: 0.0832 (0.0925)  loss: 0.8231 (0.8207)  time: 0.2753  data: 0.0003  max mem: 26157
Train: [32]  [ 700/6250]  eta: 0:25:49  lr: 0.000102  grad: 0.0824 (0.0916)  loss: 0.8208 (0.8204)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [32]  [ 800/6250]  eta: 0:25:16  lr: 0.000101  grad: 0.0816 (0.0907)  loss: 0.8186 (0.8202)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [32]  [ 900/6250]  eta: 0:24:44  lr: 0.000101  grad: 0.0836 (0.0900)  loss: 0.8179 (0.8201)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [32]  [1000/6250]  eta: 0:25:04  lr: 0.000101  grad: 0.0830 (0.0895)  loss: 0.8099 (0.8197)  time: 0.7604  data: 0.4879  max mem: 26157
Train: [32]  [1100/6250]  eta: 0:24:28  lr: 0.000101  grad: 0.0825 (0.0892)  loss: 0.8180 (0.8195)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [32]  [1200/6250]  eta: 0:23:53  lr: 0.000101  grad: 0.0854 (0.0891)  loss: 0.8133 (0.8191)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [32]  [1300/6250]  eta: 0:23:20  lr: 0.000101  grad: 0.0862 (0.0890)  loss: 0.8167 (0.8189)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [32]  [1400/6250]  eta: 0:22:47  lr: 0.000101  grad: 0.0908 (0.0890)  loss: 0.8118 (0.8185)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [32]  [1500/6250]  eta: 0:22:15  lr: 0.000101  grad: 0.0855 (0.0891)  loss: 0.8139 (0.8181)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [32]  [1600/6250]  eta: 0:21:43  lr: 0.000101  grad: 0.0900 (0.0894)  loss: 0.7993 (0.8175)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [32]  [1700/6250]  eta: 0:21:12  lr: 0.000101  grad: 0.0824 (0.0894)  loss: 0.8111 (0.8170)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [32]  [1800/6250]  eta: 0:20:42  lr: 0.000101  grad: 0.0874 (0.0895)  loss: 0.8114 (0.8166)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [32]  [1900/6250]  eta: 0:20:11  lr: 0.000101  grad: 0.0931 (0.0897)  loss: 0.8135 (0.8164)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [32]  [2000/6250]  eta: 0:19:41  lr: 0.000101  grad: 0.0882 (0.0898)  loss: 0.8074 (0.8161)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [32]  [2100/6250]  eta: 0:19:12  lr: 0.000101  grad: 0.0884 (0.0898)  loss: 0.8140 (0.8159)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [32]  [2200/6250]  eta: 0:18:43  lr: 0.000101  grad: 0.0911 (0.0899)  loss: 0.8106 (0.8156)  time: 0.2762  data: 0.0002  max mem: 26157
Train: [32]  [2300/6250]  eta: 0:18:18  lr: 0.000101  grad: 0.0911 (0.0900)  loss: 0.8150 (0.8155)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [32]  [2400/6250]  eta: 0:17:50  lr: 0.000101  grad: 0.0866 (0.0902)  loss: 0.8103 (0.8153)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [32]  [2500/6250]  eta: 0:17:21  lr: 0.000101  grad: 0.0885 (0.0902)  loss: 0.8096 (0.8151)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [32]  [2600/6250]  eta: 0:16:52  lr: 0.000101  grad: 0.0876 (0.0904)  loss: 0.8066 (0.8147)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [32]  [2700/6250]  eta: 0:16:23  lr: 0.000101  grad: 0.0909 (0.0905)  loss: 0.8073 (0.8144)  time: 0.2690  data: 0.0003  max mem: 26157
Train: [32]  [2800/6250]  eta: 0:15:54  lr: 0.000101  grad: 0.0948 (0.0907)  loss: 0.8083 (0.8143)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [32]  [2900/6250]  eta: 0:15:26  lr: 0.000101  grad: 0.0937 (0.0907)  loss: 0.8075 (0.8141)  time: 0.2705  data: 0.0003  max mem: 26157
Train: [32]  [3000/6250]  eta: 0:14:57  lr: 0.000101  grad: 0.0864 (0.0907)  loss: 0.8130 (0.8141)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [32]  [3100/6250]  eta: 0:14:29  lr: 0.000101  grad: 0.0901 (0.0907)  loss: 0.8088 (0.8139)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [32]  [3200/6250]  eta: 0:14:01  lr: 0.000101  grad: 0.0903 (0.0908)  loss: 0.8170 (0.8138)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [32]  [3300/6250]  eta: 0:13:32  lr: 0.000101  grad: 0.0894 (0.0909)  loss: 0.8113 (0.8136)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [32]  [3400/6250]  eta: 0:13:04  lr: 0.000101  grad: 0.0897 (0.0911)  loss: 0.8072 (0.8135)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [32]  [3500/6250]  eta: 0:12:36  lr: 0.000101  grad: 0.0901 (0.0911)  loss: 0.8084 (0.8133)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [32]  [3600/6250]  eta: 0:12:10  lr: 0.000101  grad: 0.0915 (0.0911)  loss: 0.8075 (0.8131)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [32]  [3700/6250]  eta: 0:11:42  lr: 0.000101  grad: 0.0889 (0.0913)  loss: 0.8041 (0.8130)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [32]  [3800/6250]  eta: 0:11:14  lr: 0.000101  grad: 0.0962 (0.0913)  loss: 0.8038 (0.8128)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [32]  [3900/6250]  eta: 0:10:46  lr: 0.000101  grad: 0.0906 (0.0913)  loss: 0.8038 (0.8127)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [32]  [4000/6250]  eta: 0:10:18  lr: 0.000101  grad: 0.0938 (0.0913)  loss: 0.8027 (0.8126)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [32]  [4100/6250]  eta: 0:09:50  lr: 0.000101  grad: 0.0927 (0.0914)  loss: 0.8043 (0.8125)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [32]  [4200/6250]  eta: 0:09:25  lr: 0.000101  grad: 0.0888 (0.0914)  loss: 0.8104 (0.8124)  time: 0.4731  data: 0.1966  max mem: 26157
Train: [32]  [4300/6250]  eta: 0:08:57  lr: 0.000101  grad: 0.0914 (0.0914)  loss: 0.8173 (0.8123)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [32]  [4400/6250]  eta: 0:08:29  lr: 0.000101  grad: 0.0930 (0.0914)  loss: 0.8084 (0.8122)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [32]  [4500/6250]  eta: 0:08:01  lr: 0.000101  grad: 0.0889 (0.0915)  loss: 0.8152 (0.8122)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [32]  [4600/6250]  eta: 0:07:36  lr: 0.000101  grad: 0.0912 (0.0915)  loss: 0.8111 (0.8122)  time: 0.5663  data: 0.2879  max mem: 26157
Train: [32]  [4700/6250]  eta: 0:07:12  lr: 0.000100  grad: 0.0865 (0.0916)  loss: 0.8116 (0.8121)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [32]  [4800/6250]  eta: 0:06:44  lr: 0.000100  grad: 0.0939 (0.0917)  loss: 0.8137 (0.8121)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [32]  [4900/6250]  eta: 0:06:16  lr: 0.000100  grad: 0.0892 (0.0917)  loss: 0.8171 (0.8122)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [32]  [5000/6250]  eta: 0:05:48  lr: 0.000100  grad: 0.0861 (0.0918)  loss: 0.8153 (0.8122)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [32]  [5100/6250]  eta: 0:05:20  lr: 0.000100  grad: 0.0929 (0.0918)  loss: 0.8168 (0.8122)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [32]  [5200/6250]  eta: 0:04:53  lr: 0.000100  grad: 0.0896 (0.0918)  loss: 0.8133 (0.8122)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [32]  [5300/6250]  eta: 0:04:25  lr: 0.000100  grad: 0.0869 (0.0919)  loss: 0.8133 (0.8122)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [32]  [5400/6250]  eta: 0:03:57  lr: 0.000100  grad: 0.0907 (0.0919)  loss: 0.8067 (0.8122)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [32]  [5500/6250]  eta: 0:03:29  lr: 0.000100  grad: 0.0928 (0.0920)  loss: 0.8111 (0.8122)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [32]  [5600/6250]  eta: 0:03:01  lr: 0.000100  grad: 0.0903 (0.0921)  loss: 0.8103 (0.8122)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [32]  [5700/6250]  eta: 0:02:33  lr: 0.000100  grad: 0.0889 (0.0921)  loss: 0.8111 (0.8121)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [32]  [5800/6250]  eta: 0:02:05  lr: 0.000100  grad: 0.0904 (0.0922)  loss: 0.8163 (0.8122)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [32]  [5900/6250]  eta: 0:01:37  lr: 0.000100  grad: 0.0879 (0.0921)  loss: 0.8090 (0.8121)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [32]  [6000/6250]  eta: 0:01:09  lr: 0.000100  grad: 0.0932 (0.0922)  loss: 0.8058 (0.8121)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [32]  [6100/6250]  eta: 0:00:41  lr: 0.000100  grad: 0.0895 (0.0922)  loss: 0.8161 (0.8121)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [32]  [6200/6250]  eta: 0:00:13  lr: 0.000100  grad: 0.0857 (0.0921)  loss: 0.8116 (0.8121)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [32]  [6249/6250]  eta: 0:00:00  lr: 0.000100  grad: 0.0859 (0.0921)  loss: 0.8080 (0.8121)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [32] Total time: 0:29:09 (0.2799 s / it)
Averaged stats: lr: 0.000100  grad: 0.0859 (0.0921)  loss: 0.8080 (0.8121)
Eval (hcp-train-subset): [32]  [ 0/62]  eta: 0:05:18  loss: 0.8441 (0.8441)  time: 5.1322  data: 5.0481  max mem: 26157
Eval (hcp-train-subset): [32]  [61/62]  eta: 0:00:00  loss: 0.8316 (0.8327)  time: 0.1386  data: 0.0557  max mem: 26157
Eval (hcp-train-subset): [32] Total time: 0:00:13 (0.2154 s / it)
Averaged stats (hcp-train-subset): loss: 0.8316 (0.8327)
Making plots (hcp-train-subset): example=40
Eval (hcp-val): [32]  [ 0/62]  eta: 0:05:04  loss: 0.8283 (0.8283)  time: 4.9150  data: 4.8304  max mem: 26157
Eval (hcp-val): [32]  [61/62]  eta: 0:00:00  loss: 0.8265 (0.8292)  time: 0.1417  data: 0.0589  max mem: 26157
Eval (hcp-val): [32] Total time: 0:00:13 (0.2192 s / it)
Averaged stats (hcp-val): loss: 0.8265 (0.8292)
Making plots (hcp-val): example=10
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [33]  [   0/6250]  eta: 8:57:33  lr: 0.000100  grad: 0.0820 (0.0820)  loss: 0.8649 (0.8649)  time: 5.1606  data: 4.7796  max mem: 26157
Train: [33]  [ 100/6250]  eta: 0:33:46  lr: 0.000100  grad: 0.0953 (0.1131)  loss: 0.8246 (0.8239)  time: 0.2703  data: 0.0003  max mem: 26157
Train: [33]  [ 200/6250]  eta: 0:30:16  lr: 0.000100  grad: 0.0921 (0.1082)  loss: 0.8252 (0.8201)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [33]  [ 300/6250]  eta: 0:28:49  lr: 0.000100  grad: 0.0964 (0.1055)  loss: 0.8136 (0.8167)  time: 0.2799  data: 0.0006  max mem: 26157
Train: [33]  [ 400/6250]  eta: 0:27:49  lr: 0.000100  grad: 0.0884 (0.1023)  loss: 0.8236 (0.8162)  time: 0.2699  data: 0.0001  max mem: 26157
Train: [33]  [ 500/6250]  eta: 0:27:03  lr: 0.000100  grad: 0.0862 (0.0993)  loss: 0.8199 (0.8167)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [33]  [ 600/6250]  eta: 0:26:24  lr: 0.000100  grad: 0.0927 (0.0976)  loss: 0.8191 (0.8169)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [33]  [ 700/6250]  eta: 0:25:47  lr: 0.000100  grad: 0.0860 (0.0964)  loss: 0.8162 (0.8172)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [33]  [ 800/6250]  eta: 0:25:14  lr: 0.000100  grad: 0.0838 (0.0960)  loss: 0.8201 (0.8174)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [33]  [ 900/6250]  eta: 0:24:41  lr: 0.000100  grad: 0.0852 (0.0953)  loss: 0.8156 (0.8174)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [33]  [1000/6250]  eta: 0:24:09  lr: 0.000100  grad: 0.0881 (0.0944)  loss: 0.8173 (0.8174)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [33]  [1100/6250]  eta: 0:23:38  lr: 0.000100  grad: 0.0836 (0.0940)  loss: 0.8176 (0.8172)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [33]  [1200/6250]  eta: 0:23:09  lr: 0.000100  grad: 0.0913 (0.0942)  loss: 0.8064 (0.8170)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [33]  [1300/6250]  eta: 0:22:39  lr: 0.000100  grad: 0.0918 (0.0940)  loss: 0.8126 (0.8167)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [33]  [1400/6250]  eta: 0:22:09  lr: 0.000100  grad: 0.0931 (0.0937)  loss: 0.8097 (0.8164)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [33]  [1500/6250]  eta: 0:21:41  lr: 0.000100  grad: 0.0873 (0.0938)  loss: 0.8117 (0.8159)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [33]  [1600/6250]  eta: 0:21:12  lr: 0.000100  grad: 0.0914 (0.0936)  loss: 0.8132 (0.8157)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [33]  [1700/6250]  eta: 0:20:47  lr: 0.000100  grad: 0.0909 (0.0935)  loss: 0.8116 (0.8154)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [33]  [1800/6250]  eta: 0:20:33  lr: 0.000100  grad: 0.0961 (0.0935)  loss: 0.8105 (0.8152)  time: 0.4895  data: 0.2163  max mem: 26157
Train: [33]  [1900/6250]  eta: 0:20:03  lr: 0.000100  grad: 0.0886 (0.0937)  loss: 0.8180 (0.8150)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [33]  [2000/6250]  eta: 0:19:34  lr: 0.000100  grad: 0.0938 (0.0937)  loss: 0.8092 (0.8148)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [33]  [2100/6250]  eta: 0:19:05  lr: 0.000100  grad: 0.0885 (0.0936)  loss: 0.8144 (0.8147)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [33]  [2200/6250]  eta: 0:18:36  lr: 0.000099  grad: 0.0884 (0.0935)  loss: 0.8071 (0.8144)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [33]  [2300/6250]  eta: 0:18:17  lr: 0.000099  grad: 0.1017 (0.0936)  loss: 0.8049 (0.8142)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [33]  [2400/6250]  eta: 0:17:48  lr: 0.000099  grad: 0.0946 (0.0936)  loss: 0.8083 (0.8141)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [33]  [2500/6250]  eta: 0:17:22  lr: 0.000099  grad: 0.0827 (0.0935)  loss: 0.8115 (0.8140)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [33]  [2600/6250]  eta: 0:16:54  lr: 0.000099  grad: 0.0906 (0.0936)  loss: 0.8109 (0.8139)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [33]  [2700/6250]  eta: 0:16:25  lr: 0.000099  grad: 0.0922 (0.0935)  loss: 0.8059 (0.8138)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [33]  [2800/6250]  eta: 0:16:07  lr: 0.000099  grad: 0.0941 (0.0935)  loss: 0.8036 (0.8137)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [33]  [2900/6250]  eta: 0:15:38  lr: 0.000099  grad: 0.0886 (0.0935)  loss: 0.8079 (0.8136)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [33]  [3000/6250]  eta: 0:15:09  lr: 0.000099  grad: 0.0919 (0.0936)  loss: 0.8083 (0.8135)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [33]  [3100/6250]  eta: 0:14:40  lr: 0.000099  grad: 0.0965 (0.0936)  loss: 0.8095 (0.8134)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [33]  [3200/6250]  eta: 0:14:11  lr: 0.000099  grad: 0.0955 (0.0940)  loss: 0.8101 (0.8133)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [33]  [3300/6250]  eta: 0:13:42  lr: 0.000099  grad: 0.0857 (0.0940)  loss: 0.8130 (0.8132)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [33]  [3400/6250]  eta: 0:13:13  lr: 0.000099  grad: 0.0876 (0.0940)  loss: 0.8167 (0.8131)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [33]  [3500/6250]  eta: 0:12:45  lr: 0.000099  grad: 0.0906 (0.0939)  loss: 0.8129 (0.8131)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [33]  [3600/6250]  eta: 0:12:16  lr: 0.000099  grad: 0.0944 (0.0940)  loss: 0.8071 (0.8129)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [33]  [3700/6250]  eta: 0:11:48  lr: 0.000099  grad: 0.0920 (0.0940)  loss: 0.8103 (0.8128)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [33]  [3800/6250]  eta: 0:11:23  lr: 0.000099  grad: 0.0923 (0.0939)  loss: 0.8129 (0.8127)  time: 0.3133  data: 0.0410  max mem: 26157
Train: [33]  [3900/6250]  eta: 0:10:54  lr: 0.000099  grad: 0.0921 (0.0940)  loss: 0.8086 (0.8126)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [33]  [4000/6250]  eta: 0:10:26  lr: 0.000099  grad: 0.0854 (0.0939)  loss: 0.8115 (0.8125)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [33]  [4100/6250]  eta: 0:09:58  lr: 0.000099  grad: 0.0877 (0.0941)  loss: 0.8131 (0.8124)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [33]  [4200/6250]  eta: 0:09:29  lr: 0.000099  grad: 0.0857 (0.0941)  loss: 0.8133 (0.8124)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [33]  [4300/6250]  eta: 0:09:01  lr: 0.000099  grad: 0.0903 (0.0941)  loss: 0.8063 (0.8123)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [33]  [4400/6250]  eta: 0:08:33  lr: 0.000099  grad: 0.0927 (0.0940)  loss: 0.8104 (0.8123)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [33]  [4500/6250]  eta: 0:08:05  lr: 0.000099  grad: 0.0928 (0.0941)  loss: 0.8129 (0.8123)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [33]  [4600/6250]  eta: 0:07:37  lr: 0.000099  grad: 0.0925 (0.0942)  loss: 0.8106 (0.8122)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [33]  [4700/6250]  eta: 0:07:09  lr: 0.000099  grad: 0.0895 (0.0942)  loss: 0.8095 (0.8122)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [33]  [4800/6250]  eta: 0:06:41  lr: 0.000099  grad: 0.0878 (0.0942)  loss: 0.8086 (0.8121)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [33]  [4900/6250]  eta: 0:06:13  lr: 0.000099  grad: 0.0898 (0.0943)  loss: 0.8055 (0.8120)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [33]  [5000/6250]  eta: 0:05:45  lr: 0.000099  grad: 0.0954 (0.0943)  loss: 0.8056 (0.8120)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [33]  [5100/6250]  eta: 0:05:17  lr: 0.000099  grad: 0.0890 (0.0942)  loss: 0.8134 (0.8120)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [33]  [5200/6250]  eta: 0:04:50  lr: 0.000099  grad: 0.0866 (0.0942)  loss: 0.8095 (0.8119)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [33]  [5300/6250]  eta: 0:04:22  lr: 0.000099  grad: 0.0925 (0.0943)  loss: 0.8010 (0.8119)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [33]  [5400/6250]  eta: 0:03:54  lr: 0.000099  grad: 0.0943 (0.0943)  loss: 0.8108 (0.8118)  time: 0.2775  data: 0.0002  max mem: 26157
Train: [33]  [5500/6250]  eta: 0:03:27  lr: 0.000099  grad: 0.0894 (0.0943)  loss: 0.8116 (0.8118)  time: 0.4104  data: 0.1330  max mem: 26157
Train: [33]  [5600/6250]  eta: 0:02:59  lr: 0.000099  grad: 0.0937 (0.0943)  loss: 0.8112 (0.8117)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [33]  [5700/6250]  eta: 0:02:32  lr: 0.000099  grad: 0.0936 (0.0943)  loss: 0.8087 (0.8117)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [33]  [5800/6250]  eta: 0:02:04  lr: 0.000099  grad: 0.0952 (0.0944)  loss: 0.8011 (0.8115)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [33]  [5900/6250]  eta: 0:01:36  lr: 0.000098  grad: 0.0913 (0.0944)  loss: 0.8073 (0.8115)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [33]  [6000/6250]  eta: 0:01:09  lr: 0.000098  grad: 0.0888 (0.0944)  loss: 0.8134 (0.8115)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [33]  [6100/6250]  eta: 0:00:41  lr: 0.000098  grad: 0.0902 (0.0945)  loss: 0.8080 (0.8114)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [33]  [6200/6250]  eta: 0:00:13  lr: 0.000098  grad: 0.0889 (0.0944)  loss: 0.8164 (0.8114)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [33]  [6249/6250]  eta: 0:00:00  lr: 0.000098  grad: 0.0889 (0.0944)  loss: 0.8102 (0.8114)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [33] Total time: 0:28:52 (0.2771 s / it)
Averaged stats: lr: 0.000098  grad: 0.0889 (0.0944)  loss: 0.8102 (0.8114)
Eval (hcp-train-subset): [33]  [ 0/62]  eta: 0:06:08  loss: 0.8435 (0.8435)  time: 5.9364  data: 5.8518  max mem: 26157
Eval (hcp-train-subset): [33]  [61/62]  eta: 0:00:00  loss: 0.8254 (0.8282)  time: 0.1295  data: 0.0447  max mem: 26157
Eval (hcp-train-subset): [33] Total time: 0:00:13 (0.2143 s / it)
Averaged stats (hcp-train-subset): loss: 0.8254 (0.8282)
Making plots (hcp-train-subset): example=47
Eval (hcp-val): [33]  [ 0/62]  eta: 0:03:39  loss: 0.8238 (0.8238)  time: 3.5374  data: 3.4269  max mem: 26157
Eval (hcp-val): [33]  [61/62]  eta: 0:00:00  loss: 0.8274 (0.8283)  time: 0.1324  data: 0.0489  max mem: 26157
Eval (hcp-val): [33] Total time: 0:00:12 (0.2074 s / it)
Averaged stats (hcp-val): loss: 0.8274 (0.8283)
Making plots (hcp-val): example=26
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [34]  [   0/6250]  eta: 12:26:52  lr: 0.000098  grad: 0.1531 (0.1531)  loss: 0.7917 (0.7917)  time: 7.1700  data: 6.8870  max mem: 26157
Train: [34]  [ 100/6250]  eta: 0:34:53  lr: 0.000098  grad: 0.0866 (0.0906)  loss: 0.8311 (0.8316)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [34]  [ 200/6250]  eta: 0:30:49  lr: 0.000098  grad: 0.0896 (0.0894)  loss: 0.8195 (0.8274)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [34]  [ 300/6250]  eta: 0:29:07  lr: 0.000098  grad: 0.0869 (0.0897)  loss: 0.8247 (0.8244)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [34]  [ 400/6250]  eta: 0:28:08  lr: 0.000098  grad: 0.0864 (0.0894)  loss: 0.8162 (0.8230)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [34]  [ 500/6250]  eta: 0:27:17  lr: 0.000098  grad: 0.0881 (0.0890)  loss: 0.8170 (0.8220)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [34]  [ 600/6250]  eta: 0:26:35  lr: 0.000098  grad: 0.0802 (0.0884)  loss: 0.8226 (0.8216)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [34]  [ 700/6250]  eta: 0:25:58  lr: 0.000098  grad: 0.0793 (0.0877)  loss: 0.8184 (0.8215)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [34]  [ 800/6250]  eta: 0:25:22  lr: 0.000098  grad: 0.0787 (0.0872)  loss: 0.8233 (0.8216)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [34]  [ 900/6250]  eta: 0:24:49  lr: 0.000098  grad: 0.0848 (0.0871)  loss: 0.8160 (0.8214)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [34]  [1000/6250]  eta: 0:24:43  lr: 0.000098  grad: 0.0868 (0.0871)  loss: 0.8256 (0.8213)  time: 0.5099  data: 0.2354  max mem: 26157
Train: [34]  [1100/6250]  eta: 0:24:16  lr: 0.000098  grad: 0.0815 (0.0870)  loss: 0.8190 (0.8209)  time: 0.3351  data: 0.0641  max mem: 26157
Train: [34]  [1200/6250]  eta: 0:23:50  lr: 0.000098  grad: 0.0795 (0.0870)  loss: 0.8152 (0.8205)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [34]  [1300/6250]  eta: 0:23:17  lr: 0.000098  grad: 0.0815 (0.0870)  loss: 0.8175 (0.8202)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [34]  [1400/6250]  eta: 0:22:52  lr: 0.000098  grad: 0.0866 (0.0872)  loss: 0.8182 (0.8199)  time: 0.3804  data: 0.1085  max mem: 26157
Train: [34]  [1500/6250]  eta: 0:22:20  lr: 0.000098  grad: 0.0847 (0.0872)  loss: 0.8075 (0.8195)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [34]  [1600/6250]  eta: 0:21:48  lr: 0.000098  grad: 0.0868 (0.0874)  loss: 0.8078 (0.8189)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [34]  [1700/6250]  eta: 0:21:16  lr: 0.000098  grad: 0.0897 (0.0876)  loss: 0.8128 (0.8185)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [34]  [1800/6250]  eta: 0:20:46  lr: 0.000098  grad: 0.0935 (0.0878)  loss: 0.8110 (0.8181)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [34]  [1900/6250]  eta: 0:20:16  lr: 0.000098  grad: 0.0945 (0.0883)  loss: 0.8100 (0.8176)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [34]  [2000/6250]  eta: 0:19:45  lr: 0.000098  grad: 0.0878 (0.0886)  loss: 0.8068 (0.8172)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [34]  [2100/6250]  eta: 0:19:16  lr: 0.000098  grad: 0.0958 (0.0889)  loss: 0.8110 (0.8167)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [34]  [2200/6250]  eta: 0:18:47  lr: 0.000098  grad: 0.0935 (0.0893)  loss: 0.8098 (0.8163)  time: 0.2746  data: 0.0002  max mem: 26157
Train: [34]  [2300/6250]  eta: 0:18:18  lr: 0.000098  grad: 0.0912 (0.0894)  loss: 0.8037 (0.8160)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [34]  [2400/6250]  eta: 0:17:48  lr: 0.000098  grad: 0.0934 (0.0897)  loss: 0.8097 (0.8156)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [34]  [2500/6250]  eta: 0:17:20  lr: 0.000098  grad: 0.0899 (0.0899)  loss: 0.8086 (0.8153)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [34]  [2600/6250]  eta: 0:16:51  lr: 0.000098  grad: 0.0967 (0.0902)  loss: 0.8008 (0.8149)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [34]  [2700/6250]  eta: 0:16:22  lr: 0.000098  grad: 0.0947 (0.0904)  loss: 0.8040 (0.8147)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [34]  [2800/6250]  eta: 0:15:54  lr: 0.000098  grad: 0.0937 (0.0906)  loss: 0.8092 (0.8144)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [34]  [2900/6250]  eta: 0:15:25  lr: 0.000098  grad: 0.0958 (0.0909)  loss: 0.8040 (0.8142)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [34]  [3000/6250]  eta: 0:14:57  lr: 0.000098  grad: 0.0859 (0.0911)  loss: 0.8157 (0.8141)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [34]  [3100/6250]  eta: 0:14:29  lr: 0.000098  grad: 0.0950 (0.0912)  loss: 0.8103 (0.8140)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [34]  [3200/6250]  eta: 0:14:00  lr: 0.000098  grad: 0.0892 (0.0913)  loss: 0.8112 (0.8139)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [34]  [3300/6250]  eta: 0:13:33  lr: 0.000097  grad: 0.0941 (0.0914)  loss: 0.8076 (0.8138)  time: 0.2863  data: 0.0167  max mem: 26157
Train: [34]  [3400/6250]  eta: 0:13:04  lr: 0.000097  grad: 0.0967 (0.0916)  loss: 0.8089 (0.8137)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [34]  [3500/6250]  eta: 0:12:38  lr: 0.000097  grad: 0.0911 (0.0916)  loss: 0.8152 (0.8136)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [34]  [3600/6250]  eta: 0:12:10  lr: 0.000097  grad: 0.0920 (0.0918)  loss: 0.8081 (0.8135)  time: 0.2869  data: 0.0126  max mem: 26157
Train: [34]  [3700/6250]  eta: 0:11:42  lr: 0.000097  grad: 0.0922 (0.0919)  loss: 0.8079 (0.8135)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [34]  [3800/6250]  eta: 0:11:14  lr: 0.000097  grad: 0.0876 (0.0921)  loss: 0.8132 (0.8135)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [34]  [3900/6250]  eta: 0:10:50  lr: 0.000097  grad: 0.0937 (0.0923)  loss: 0.8096 (0.8134)  time: 0.5927  data: 0.3224  max mem: 26157
Train: [34]  [4000/6250]  eta: 0:10:22  lr: 0.000097  grad: 0.0921 (0.0925)  loss: 0.8116 (0.8134)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [34]  [4100/6250]  eta: 0:09:54  lr: 0.000097  grad: 0.0880 (0.0925)  loss: 0.8146 (0.8134)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [34]  [4200/6250]  eta: 0:09:26  lr: 0.000097  grad: 0.0968 (0.0925)  loss: 0.8097 (0.8134)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [34]  [4300/6250]  eta: 0:08:58  lr: 0.000097  grad: 0.0965 (0.0926)  loss: 0.8135 (0.8134)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [34]  [4400/6250]  eta: 0:08:30  lr: 0.000097  grad: 0.1024 (0.0927)  loss: 0.8119 (0.8133)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [34]  [4500/6250]  eta: 0:08:02  lr: 0.000097  grad: 0.0983 (0.0930)  loss: 0.8036 (0.8133)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [34]  [4600/6250]  eta: 0:07:35  lr: 0.000097  grad: 0.0898 (0.0931)  loss: 0.8143 (0.8132)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [34]  [4700/6250]  eta: 0:07:07  lr: 0.000097  grad: 0.0955 (0.0931)  loss: 0.8138 (0.8132)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [34]  [4800/6250]  eta: 0:06:39  lr: 0.000097  grad: 0.0918 (0.0931)  loss: 0.8140 (0.8132)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [34]  [4900/6250]  eta: 0:06:11  lr: 0.000097  grad: 0.0917 (0.0931)  loss: 0.8141 (0.8132)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [34]  [5000/6250]  eta: 0:05:44  lr: 0.000097  grad: 0.0930 (0.0932)  loss: 0.8115 (0.8133)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [34]  [5100/6250]  eta: 0:05:16  lr: 0.000097  grad: 0.0931 (0.0932)  loss: 0.8156 (0.8133)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [34]  [5200/6250]  eta: 0:04:49  lr: 0.000097  grad: 0.0898 (0.0932)  loss: 0.8192 (0.8134)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [34]  [5300/6250]  eta: 0:04:21  lr: 0.000097  grad: 0.0953 (0.0932)  loss: 0.8148 (0.8134)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [34]  [5400/6250]  eta: 0:03:54  lr: 0.000097  grad: 0.0929 (0.0932)  loss: 0.8162 (0.8134)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [34]  [5500/6250]  eta: 0:03:26  lr: 0.000097  grad: 0.0898 (0.0933)  loss: 0.8160 (0.8135)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [34]  [5600/6250]  eta: 0:02:58  lr: 0.000097  grad: 0.0868 (0.0934)  loss: 0.8159 (0.8135)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [34]  [5700/6250]  eta: 0:02:31  lr: 0.000097  grad: 0.0837 (0.0934)  loss: 0.8150 (0.8135)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [34]  [5800/6250]  eta: 0:02:03  lr: 0.000097  grad: 0.0883 (0.0934)  loss: 0.8095 (0.8135)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [34]  [5900/6250]  eta: 0:01:36  lr: 0.000097  grad: 0.0885 (0.0934)  loss: 0.8144 (0.8135)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [34]  [6000/6250]  eta: 0:01:08  lr: 0.000097  grad: 0.0947 (0.0934)  loss: 0.8179 (0.8136)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [34]  [6100/6250]  eta: 0:00:41  lr: 0.000097  grad: 0.0925 (0.0934)  loss: 0.8173 (0.8136)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [34]  [6200/6250]  eta: 0:00:13  lr: 0.000097  grad: 0.0990 (0.0934)  loss: 0.8121 (0.8136)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [34]  [6249/6250]  eta: 0:00:00  lr: 0.000097  grad: 0.0934 (0.0934)  loss: 0.8127 (0.8137)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [34] Total time: 0:28:43 (0.2757 s / it)
Averaged stats: lr: 0.000097  grad: 0.0934 (0.0934)  loss: 0.8127 (0.8137)
Eval (hcp-train-subset): [34]  [ 0/62]  eta: 0:04:52  loss: 0.8356 (0.8356)  time: 4.7236  data: 4.6390  max mem: 26157
Eval (hcp-train-subset): [34]  [61/62]  eta: 0:00:00  loss: 0.8303 (0.8310)  time: 0.1238  data: 0.0411  max mem: 26157
Eval (hcp-train-subset): [34] Total time: 0:00:12 (0.2080 s / it)
Averaged stats (hcp-train-subset): loss: 0.8303 (0.8310)
Making plots (hcp-train-subset): example=9
Eval (hcp-val): [34]  [ 0/62]  eta: 0:05:32  loss: 0.8221 (0.8221)  time: 5.3688  data: 5.2847  max mem: 26157
Eval (hcp-val): [34]  [61/62]  eta: 0:00:00  loss: 0.8280 (0.8284)  time: 0.1291  data: 0.0438  max mem: 26157
Eval (hcp-val): [34] Total time: 0:00:13 (0.2148 s / it)
Averaged stats (hcp-val): loss: 0.8280 (0.8284)
Making plots (hcp-val): example=60
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [35]  [   0/6250]  eta: 7:21:51  lr: 0.000097  grad: 0.0647 (0.0647)  loss: 0.8761 (0.8761)  time: 4.2418  data: 3.8645  max mem: 26157
Train: [35]  [ 100/6250]  eta: 0:35:24  lr: 0.000097  grad: 0.0934 (0.1045)  loss: 0.8261 (0.8268)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [35]  [ 200/6250]  eta: 0:31:02  lr: 0.000097  grad: 0.0974 (0.1022)  loss: 0.8129 (0.8202)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [35]  [ 300/6250]  eta: 0:29:15  lr: 0.000097  grad: 0.0954 (0.1008)  loss: 0.8104 (0.8179)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [35]  [ 400/6250]  eta: 0:28:10  lr: 0.000097  grad: 0.0942 (0.0988)  loss: 0.8118 (0.8168)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [35]  [ 500/6250]  eta: 0:27:21  lr: 0.000097  grad: 0.0863 (0.0977)  loss: 0.8134 (0.8162)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [35]  [ 600/6250]  eta: 0:26:38  lr: 0.000097  grad: 0.0896 (0.0967)  loss: 0.8119 (0.8160)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [35]  [ 700/6250]  eta: 0:25:58  lr: 0.000096  grad: 0.0913 (0.0966)  loss: 0.8139 (0.8154)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [35]  [ 800/6250]  eta: 0:25:23  lr: 0.000096  grad: 0.0872 (0.0962)  loss: 0.8190 (0.8148)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [35]  [ 900/6250]  eta: 0:24:49  lr: 0.000096  grad: 0.0932 (0.0961)  loss: 0.8082 (0.8144)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [35]  [1000/6250]  eta: 0:24:17  lr: 0.000096  grad: 0.0906 (0.0959)  loss: 0.8140 (0.8141)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [35]  [1100/6250]  eta: 0:23:52  lr: 0.000096  grad: 0.0894 (0.0957)  loss: 0.8116 (0.8137)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [35]  [1200/6250]  eta: 0:23:43  lr: 0.000096  grad: 0.0866 (0.0953)  loss: 0.8070 (0.8133)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [35]  [1300/6250]  eta: 0:23:09  lr: 0.000096  grad: 0.0943 (0.0952)  loss: 0.8094 (0.8128)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [35]  [1400/6250]  eta: 0:22:38  lr: 0.000096  grad: 0.0924 (0.0951)  loss: 0.8112 (0.8126)  time: 0.2721  data: 0.0003  max mem: 26157
Train: [35]  [1500/6250]  eta: 0:22:16  lr: 0.000096  grad: 0.0975 (0.0954)  loss: 0.8113 (0.8123)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [35]  [1600/6250]  eta: 0:21:47  lr: 0.000096  grad: 0.0976 (0.0959)  loss: 0.8054 (0.8120)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [35]  [1700/6250]  eta: 0:21:20  lr: 0.000096  grad: 0.0906 (0.0957)  loss: 0.8172 (0.8119)  time: 0.3516  data: 0.0814  max mem: 26157
Train: [35]  [1800/6250]  eta: 0:20:53  lr: 0.000096  grad: 0.0864 (0.0957)  loss: 0.8086 (0.8118)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [35]  [1900/6250]  eta: 0:20:22  lr: 0.000096  grad: 0.0926 (0.0955)  loss: 0.8063 (0.8116)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [35]  [2000/6250]  eta: 0:19:51  lr: 0.000096  grad: 0.0872 (0.0955)  loss: 0.8124 (0.8116)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [35]  [2100/6250]  eta: 0:19:21  lr: 0.000096  grad: 0.0941 (0.0955)  loss: 0.8180 (0.8115)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [35]  [2200/6250]  eta: 0:18:51  lr: 0.000096  grad: 0.0934 (0.0955)  loss: 0.8137 (0.8114)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [35]  [2300/6250]  eta: 0:18:21  lr: 0.000096  grad: 0.0879 (0.0954)  loss: 0.8139 (0.8114)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [35]  [2400/6250]  eta: 0:17:52  lr: 0.000096  grad: 0.0916 (0.0953)  loss: 0.8072 (0.8113)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [35]  [2500/6250]  eta: 0:17:23  lr: 0.000096  grad: 0.0919 (0.0952)  loss: 0.8102 (0.8112)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [35]  [2600/6250]  eta: 0:16:54  lr: 0.000096  grad: 0.0944 (0.0952)  loss: 0.8074 (0.8110)  time: 0.2725  data: 0.0003  max mem: 26157
Train: [35]  [2700/6250]  eta: 0:16:25  lr: 0.000096  grad: 0.0947 (0.0953)  loss: 0.8048 (0.8109)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [35]  [2800/6250]  eta: 0:15:56  lr: 0.000096  grad: 0.0922 (0.0953)  loss: 0.8080 (0.8109)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [35]  [2900/6250]  eta: 0:15:28  lr: 0.000096  grad: 0.0960 (0.0953)  loss: 0.8129 (0.8109)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [35]  [3000/6250]  eta: 0:14:59  lr: 0.000096  grad: 0.0940 (0.0953)  loss: 0.8089 (0.8108)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [35]  [3100/6250]  eta: 0:14:31  lr: 0.000096  grad: 0.0912 (0.0953)  loss: 0.8112 (0.8108)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [35]  [3200/6250]  eta: 0:14:02  lr: 0.000096  grad: 0.0950 (0.0953)  loss: 0.8029 (0.8107)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [35]  [3300/6250]  eta: 0:13:34  lr: 0.000096  grad: 0.0912 (0.0952)  loss: 0.8134 (0.8108)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [35]  [3400/6250]  eta: 0:13:06  lr: 0.000096  grad: 0.0969 (0.0953)  loss: 0.8072 (0.8108)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [35]  [3500/6250]  eta: 0:12:38  lr: 0.000096  grad: 0.0849 (0.0952)  loss: 0.8128 (0.8108)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [35]  [3600/6250]  eta: 0:12:10  lr: 0.000096  grad: 0.0988 (0.0952)  loss: 0.8134 (0.8108)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [35]  [3700/6250]  eta: 0:11:42  lr: 0.000096  grad: 0.0921 (0.0952)  loss: 0.8092 (0.8108)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [35]  [3800/6250]  eta: 0:11:14  lr: 0.000096  grad: 0.0957 (0.0952)  loss: 0.8033 (0.8108)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [35]  [3900/6250]  eta: 0:10:46  lr: 0.000096  grad: 0.0894 (0.0952)  loss: 0.8122 (0.8107)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [35]  [4000/6250]  eta: 0:10:18  lr: 0.000096  grad: 0.0924 (0.0952)  loss: 0.8095 (0.8107)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [35]  [4100/6250]  eta: 0:09:50  lr: 0.000096  grad: 0.0946 (0.0952)  loss: 0.8099 (0.8106)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [35]  [4200/6250]  eta: 0:09:23  lr: 0.000096  grad: 0.0910 (0.0952)  loss: 0.8090 (0.8106)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [35]  [4300/6250]  eta: 0:08:55  lr: 0.000095  grad: 0.0902 (0.0952)  loss: 0.8126 (0.8105)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [35]  [4400/6250]  eta: 0:08:27  lr: 0.000095  grad: 0.0909 (0.0953)  loss: 0.8064 (0.8105)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [35]  [4500/6250]  eta: 0:08:00  lr: 0.000095  grad: 0.1054 (0.0954)  loss: 0.8063 (0.8104)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [35]  [4600/6250]  eta: 0:07:36  lr: 0.000095  grad: 0.0894 (0.0954)  loss: 0.8127 (0.8103)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [35]  [4700/6250]  eta: 0:07:09  lr: 0.000095  grad: 0.0943 (0.0954)  loss: 0.8064 (0.8103)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [35]  [4800/6250]  eta: 0:06:42  lr: 0.000095  grad: 0.0945 (0.0954)  loss: 0.8092 (0.8102)  time: 0.5043  data: 0.2308  max mem: 26157
Train: [35]  [4900/6250]  eta: 0:06:14  lr: 0.000095  grad: 0.0945 (0.0956)  loss: 0.8059 (0.8102)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [35]  [5000/6250]  eta: 0:05:46  lr: 0.000095  grad: 0.0926 (0.0955)  loss: 0.8074 (0.8102)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [35]  [5100/6250]  eta: 0:05:18  lr: 0.000095  grad: 0.0917 (0.0955)  loss: 0.8174 (0.8102)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [35]  [5200/6250]  eta: 0:04:50  lr: 0.000095  grad: 0.0873 (0.0955)  loss: 0.8123 (0.8102)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [35]  [5300/6250]  eta: 0:04:23  lr: 0.000095  grad: 0.0921 (0.0955)  loss: 0.8109 (0.8102)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [35]  [5400/6250]  eta: 0:03:55  lr: 0.000095  grad: 0.0938 (0.0955)  loss: 0.8100 (0.8102)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [35]  [5500/6250]  eta: 0:03:28  lr: 0.000095  grad: 0.0889 (0.0955)  loss: 0.8144 (0.8102)  time: 0.6104  data: 0.3354  max mem: 26157
Train: [35]  [5600/6250]  eta: 0:03:00  lr: 0.000095  grad: 0.0895 (0.0956)  loss: 0.8146 (0.8102)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [35]  [5700/6250]  eta: 0:02:32  lr: 0.000095  grad: 0.0838 (0.0956)  loss: 0.8116 (0.8101)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [35]  [5800/6250]  eta: 0:02:05  lr: 0.000095  grad: 0.0929 (0.0955)  loss: 0.8128 (0.8101)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [35]  [5900/6250]  eta: 0:01:37  lr: 0.000095  grad: 0.0896 (0.0955)  loss: 0.8119 (0.8101)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [35]  [6000/6250]  eta: 0:01:09  lr: 0.000095  grad: 0.0928 (0.0954)  loss: 0.8096 (0.8101)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [35]  [6100/6250]  eta: 0:00:41  lr: 0.000095  grad: 0.0859 (0.0954)  loss: 0.8138 (0.8101)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [35]  [6200/6250]  eta: 0:00:13  lr: 0.000095  grad: 0.0955 (0.0954)  loss: 0.8074 (0.8101)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [35]  [6249/6250]  eta: 0:00:00  lr: 0.000095  grad: 0.0937 (0.0954)  loss: 0.8186 (0.8101)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [35] Total time: 0:29:00 (0.2785 s / it)
Averaged stats: lr: 0.000095  grad: 0.0937 (0.0954)  loss: 0.8186 (0.8101)
Eval (hcp-train-subset): [35]  [ 0/62]  eta: 0:03:56  loss: 0.8401 (0.8401)  time: 3.8113  data: 3.6975  max mem: 26157
Eval (hcp-train-subset): [35]  [61/62]  eta: 0:00:00  loss: 0.8277 (0.8288)  time: 0.1243  data: 0.0413  max mem: 26157
Eval (hcp-train-subset): [35] Total time: 0:00:13 (0.2101 s / it)
Averaged stats (hcp-train-subset): loss: 0.8277 (0.8288)
Making plots (hcp-train-subset): example=35
Eval (hcp-val): [35]  [ 0/62]  eta: 0:04:55  loss: 0.8248 (0.8248)  time: 4.7674  data: 4.6816  max mem: 26157
Eval (hcp-val): [35]  [61/62]  eta: 0:00:00  loss: 0.8273 (0.8278)  time: 0.1305  data: 0.0479  max mem: 26157
Eval (hcp-val): [35] Total time: 0:00:13 (0.2162 s / it)
Averaged stats (hcp-val): loss: 0.8273 (0.8278)
Making plots (hcp-val): example=35
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [36]  [   0/6250]  eta: 10:33:47  lr: 0.000095  grad: 0.0796 (0.0796)  loss: 0.8487 (0.8487)  time: 6.0844  data: 5.7880  max mem: 26157
Train: [36]  [ 100/6250]  eta: 0:33:36  lr: 0.000095  grad: 0.0963 (0.1008)  loss: 0.8198 (0.8282)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [36]  [ 200/6250]  eta: 0:30:09  lr: 0.000095  grad: 0.0983 (0.1061)  loss: 0.8123 (0.8219)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [36]  [ 300/6250]  eta: 0:28:40  lr: 0.000095  grad: 0.0922 (0.1043)  loss: 0.8032 (0.8181)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [36]  [ 400/6250]  eta: 0:27:43  lr: 0.000095  grad: 0.0929 (0.1021)  loss: 0.8072 (0.8162)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [36]  [ 500/6250]  eta: 0:26:58  lr: 0.000095  grad: 0.0905 (0.1002)  loss: 0.8135 (0.8151)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [36]  [ 600/6250]  eta: 0:26:21  lr: 0.000095  grad: 0.0876 (0.0986)  loss: 0.8166 (0.8148)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [36]  [ 700/6250]  eta: 0:25:46  lr: 0.000095  grad: 0.0861 (0.0974)  loss: 0.8129 (0.8145)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [36]  [ 800/6250]  eta: 0:25:13  lr: 0.000095  grad: 0.0902 (0.0966)  loss: 0.8084 (0.8143)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [36]  [ 900/6250]  eta: 0:24:54  lr: 0.000095  grad: 0.0903 (0.0960)  loss: 0.8190 (0.8145)  time: 0.2750  data: 0.0002  max mem: 26157
Train: [36]  [1000/6250]  eta: 0:24:42  lr: 0.000095  grad: 0.0877 (0.0955)  loss: 0.8121 (0.8146)  time: 0.2742  data: 0.0003  max mem: 26157
Train: [36]  [1100/6250]  eta: 0:24:09  lr: 0.000095  grad: 0.0933 (0.0952)  loss: 0.8082 (0.8142)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [36]  [1200/6250]  eta: 0:23:36  lr: 0.000095  grad: 0.0893 (0.0953)  loss: 0.8104 (0.8140)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [36]  [1300/6250]  eta: 0:23:04  lr: 0.000095  grad: 0.0894 (0.0951)  loss: 0.8141 (0.8138)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [36]  [1400/6250]  eta: 0:22:33  lr: 0.000095  grad: 0.0920 (0.0951)  loss: 0.8102 (0.8136)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [36]  [1500/6250]  eta: 0:22:02  lr: 0.000095  grad: 0.0938 (0.0952)  loss: 0.8084 (0.8134)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [36]  [1600/6250]  eta: 0:21:32  lr: 0.000094  grad: 0.0960 (0.0951)  loss: 0.8104 (0.8132)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [36]  [1700/6250]  eta: 0:21:02  lr: 0.000094  grad: 0.0892 (0.0950)  loss: 0.8148 (0.8132)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [36]  [1800/6250]  eta: 0:20:32  lr: 0.000094  grad: 0.0895 (0.0951)  loss: 0.8217 (0.8131)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [36]  [1900/6250]  eta: 0:20:03  lr: 0.000094  grad: 0.0935 (0.0949)  loss: 0.8153 (0.8130)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [36]  [2000/6250]  eta: 0:19:34  lr: 0.000094  grad: 0.0906 (0.0949)  loss: 0.8153 (0.8130)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [36]  [2100/6250]  eta: 0:19:05  lr: 0.000094  grad: 0.0942 (0.0949)  loss: 0.8076 (0.8129)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [36]  [2200/6250]  eta: 0:18:36  lr: 0.000094  grad: 0.0878 (0.0949)  loss: 0.8111 (0.8129)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [36]  [2300/6250]  eta: 0:18:15  lr: 0.000094  grad: 0.0904 (0.0948)  loss: 0.8156 (0.8129)  time: 0.4894  data: 0.2189  max mem: 26157
Train: [36]  [2400/6250]  eta: 0:17:46  lr: 0.000094  grad: 0.0968 (0.0948)  loss: 0.8137 (0.8129)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [36]  [2500/6250]  eta: 0:17:17  lr: 0.000094  grad: 0.0912 (0.0947)  loss: 0.8171 (0.8129)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [36]  [2600/6250]  eta: 0:17:03  lr: 0.000094  grad: 0.0912 (0.0947)  loss: 0.8136 (0.8129)  time: 0.3798  data: 0.1021  max mem: 26157
Train: [36]  [2700/6250]  eta: 0:16:40  lr: 0.000094  grad: 0.0881 (0.0947)  loss: 0.8135 (0.8129)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [36]  [2800/6250]  eta: 0:16:10  lr: 0.000094  grad: 0.0933 (0.0947)  loss: 0.8150 (0.8129)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [36]  [2900/6250]  eta: 0:15:41  lr: 0.000094  grad: 0.0987 (0.0949)  loss: 0.8110 (0.8129)  time: 0.2739  data: 0.0003  max mem: 26157
Train: [36]  [3000/6250]  eta: 0:15:12  lr: 0.000094  grad: 0.0932 (0.0949)  loss: 0.8099 (0.8128)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [36]  [3100/6250]  eta: 0:14:44  lr: 0.000094  grad: 0.0980 (0.0949)  loss: 0.8080 (0.8127)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [36]  [3200/6250]  eta: 0:14:15  lr: 0.000094  grad: 0.0926 (0.0949)  loss: 0.8048 (0.8127)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [36]  [3300/6250]  eta: 0:13:46  lr: 0.000094  grad: 0.0934 (0.0949)  loss: 0.8087 (0.8127)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [36]  [3400/6250]  eta: 0:13:17  lr: 0.000094  grad: 0.0913 (0.0950)  loss: 0.8144 (0.8126)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [36]  [3500/6250]  eta: 0:12:48  lr: 0.000094  grad: 0.0922 (0.0951)  loss: 0.8124 (0.8126)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [36]  [3600/6250]  eta: 0:12:19  lr: 0.000094  grad: 0.0934 (0.0951)  loss: 0.8054 (0.8125)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [36]  [3700/6250]  eta: 0:11:51  lr: 0.000094  grad: 0.0920 (0.0952)  loss: 0.8151 (0.8125)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [36]  [3800/6250]  eta: 0:11:22  lr: 0.000094  grad: 0.0990 (0.0953)  loss: 0.8124 (0.8124)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [36]  [3900/6250]  eta: 0:10:54  lr: 0.000094  grad: 0.0955 (0.0953)  loss: 0.8071 (0.8124)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [36]  [4000/6250]  eta: 0:10:26  lr: 0.000094  grad: 0.0914 (0.0953)  loss: 0.8156 (0.8124)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [36]  [4100/6250]  eta: 0:09:57  lr: 0.000094  grad: 0.0941 (0.0954)  loss: 0.8167 (0.8124)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [36]  [4200/6250]  eta: 0:09:29  lr: 0.000094  grad: 0.0931 (0.0954)  loss: 0.8182 (0.8124)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [36]  [4300/6250]  eta: 0:09:01  lr: 0.000094  grad: 0.0926 (0.0955)  loss: 0.8193 (0.8124)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [36]  [4400/6250]  eta: 0:08:33  lr: 0.000094  grad: 0.0931 (0.0955)  loss: 0.8071 (0.8123)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [36]  [4500/6250]  eta: 0:08:05  lr: 0.000094  grad: 0.0939 (0.0955)  loss: 0.8147 (0.8123)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [36]  [4600/6250]  eta: 0:07:37  lr: 0.000094  grad: 0.0917 (0.0955)  loss: 0.8135 (0.8123)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [36]  [4700/6250]  eta: 0:07:09  lr: 0.000094  grad: 0.0980 (0.0956)  loss: 0.8100 (0.8123)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [36]  [4800/6250]  eta: 0:06:41  lr: 0.000094  grad: 0.0890 (0.0956)  loss: 0.8129 (0.8123)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [36]  [4900/6250]  eta: 0:06:13  lr: 0.000094  grad: 0.0988 (0.0958)  loss: 0.8163 (0.8123)  time: 0.2723  data: 0.0003  max mem: 26157
Train: [36]  [5000/6250]  eta: 0:05:45  lr: 0.000094  grad: 0.0898 (0.0958)  loss: 0.8170 (0.8123)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [36]  [5100/6250]  eta: 0:05:18  lr: 0.000093  grad: 0.0999 (0.0958)  loss: 0.8073 (0.8123)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [36]  [5200/6250]  eta: 0:04:50  lr: 0.000093  grad: 0.0894 (0.0959)  loss: 0.8117 (0.8122)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [36]  [5300/6250]  eta: 0:04:22  lr: 0.000093  grad: 0.0956 (0.0958)  loss: 0.8122 (0.8121)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [36]  [5400/6250]  eta: 0:03:54  lr: 0.000093  grad: 0.0899 (0.0959)  loss: 0.8076 (0.8121)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [36]  [5500/6250]  eta: 0:03:27  lr: 0.000093  grad: 0.0963 (0.0959)  loss: 0.8059 (0.8121)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [36]  [5600/6250]  eta: 0:02:59  lr: 0.000093  grad: 0.0915 (0.0959)  loss: 0.8140 (0.8120)  time: 0.2837  data: 0.0002  max mem: 26157
Train: [36]  [5700/6250]  eta: 0:02:33  lr: 0.000093  grad: 0.0900 (0.0959)  loss: 0.8124 (0.8120)  time: 0.8372  data: 0.5623  max mem: 26157
Train: [36]  [5800/6250]  eta: 0:02:05  lr: 0.000093  grad: 0.0935 (0.0959)  loss: 0.8152 (0.8120)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [36]  [5900/6250]  eta: 0:01:37  lr: 0.000093  grad: 0.0930 (0.0959)  loss: 0.8100 (0.8120)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [36]  [6000/6250]  eta: 0:01:09  lr: 0.000093  grad: 0.0946 (0.0959)  loss: 0.8168 (0.8120)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [36]  [6100/6250]  eta: 0:00:41  lr: 0.000093  grad: 0.0940 (0.0959)  loss: 0.8150 (0.8120)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [36]  [6200/6250]  eta: 0:00:13  lr: 0.000093  grad: 0.0911 (0.0960)  loss: 0.8121 (0.8120)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [36]  [6249/6250]  eta: 0:00:00  lr: 0.000093  grad: 0.0972 (0.0960)  loss: 0.8114 (0.8120)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [36] Total time: 0:29:05 (0.2792 s / it)
Averaged stats: lr: 0.000093  grad: 0.0972 (0.0960)  loss: 0.8114 (0.8120)
Eval (hcp-train-subset): [36]  [ 0/62]  eta: 0:03:56  loss: 0.8359 (0.8359)  time: 3.8095  data: 3.6922  max mem: 26157
Eval (hcp-train-subset): [36]  [61/62]  eta: 0:00:00  loss: 0.8276 (0.8291)  time: 0.1415  data: 0.0568  max mem: 26157
Eval (hcp-train-subset): [36] Total time: 0:00:13 (0.2141 s / it)
Averaged stats (hcp-train-subset): loss: 0.8276 (0.8291)
Making plots (hcp-train-subset): example=57
Eval (hcp-val): [36]  [ 0/62]  eta: 0:03:27  loss: 0.8247 (0.8247)  time: 3.3464  data: 3.2523  max mem: 26157
Eval (hcp-val): [36]  [61/62]  eta: 0:00:00  loss: 0.8268 (0.8273)  time: 0.1446  data: 0.0610  max mem: 26157
Eval (hcp-val): [36] Total time: 0:00:12 (0.2086 s / it)
Averaged stats (hcp-val): loss: 0.8268 (0.8273)
Making plots (hcp-val): example=6
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [37]  [   0/6250]  eta: 8:59:59  lr: 0.000093  grad: 0.1205 (0.1205)  loss: 0.8328 (0.8328)  time: 5.1838  data: 4.8421  max mem: 26157
Train: [37]  [ 100/6250]  eta: 0:33:20  lr: 0.000093  grad: 0.1005 (0.1123)  loss: 0.8199 (0.8260)  time: 0.2711  data: 0.0003  max mem: 26157
Train: [37]  [ 200/6250]  eta: 0:30:03  lr: 0.000093  grad: 0.0977 (0.1112)  loss: 0.8105 (0.8201)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [37]  [ 300/6250]  eta: 0:28:40  lr: 0.000093  grad: 0.0857 (0.1050)  loss: 0.8230 (0.8179)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [37]  [ 400/6250]  eta: 0:27:42  lr: 0.000093  grad: 0.0904 (0.1011)  loss: 0.8205 (0.8176)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [37]  [ 500/6250]  eta: 0:26:57  lr: 0.000093  grad: 0.0844 (0.0992)  loss: 0.8159 (0.8173)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [37]  [ 600/6250]  eta: 0:26:17  lr: 0.000093  grad: 0.0908 (0.0974)  loss: 0.8103 (0.8170)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [37]  [ 700/6250]  eta: 0:25:43  lr: 0.000093  grad: 0.0842 (0.0963)  loss: 0.8138 (0.8165)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [37]  [ 800/6250]  eta: 0:25:12  lr: 0.000093  grad: 0.0857 (0.0956)  loss: 0.8152 (0.8161)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [37]  [ 900/6250]  eta: 0:24:40  lr: 0.000093  grad: 0.0847 (0.0949)  loss: 0.8227 (0.8160)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [37]  [1000/6250]  eta: 0:24:09  lr: 0.000093  grad: 0.0910 (0.0944)  loss: 0.8070 (0.8160)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [37]  [1100/6250]  eta: 0:23:39  lr: 0.000093  grad: 0.0860 (0.0941)  loss: 0.8135 (0.8156)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [37]  [1200/6250]  eta: 0:23:09  lr: 0.000093  grad: 0.0863 (0.0939)  loss: 0.8090 (0.8154)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [37]  [1300/6250]  eta: 0:22:39  lr: 0.000093  grad: 0.0915 (0.0940)  loss: 0.8083 (0.8150)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [37]  [1400/6250]  eta: 0:22:10  lr: 0.000093  grad: 0.0949 (0.0942)  loss: 0.8122 (0.8146)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [37]  [1500/6250]  eta: 0:21:41  lr: 0.000093  grad: 0.0933 (0.0944)  loss: 0.8058 (0.8142)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [37]  [1600/6250]  eta: 0:21:12  lr: 0.000093  grad: 0.0974 (0.0947)  loss: 0.8088 (0.8138)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [37]  [1700/6250]  eta: 0:20:44  lr: 0.000093  grad: 0.0877 (0.0947)  loss: 0.8108 (0.8135)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [37]  [1800/6250]  eta: 0:20:16  lr: 0.000093  grad: 0.0936 (0.0952)  loss: 0.8007 (0.8132)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [37]  [1900/6250]  eta: 0:19:48  lr: 0.000093  grad: 0.0939 (0.0953)  loss: 0.8076 (0.8130)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [37]  [2000/6250]  eta: 0:19:20  lr: 0.000093  grad: 0.0910 (0.0953)  loss: 0.8052 (0.8128)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [37]  [2100/6250]  eta: 0:19:13  lr: 0.000093  grad: 0.0946 (0.0953)  loss: 0.8144 (0.8127)  time: 0.3424  data: 0.0689  max mem: 26157
Train: [37]  [2200/6250]  eta: 0:18:47  lr: 0.000093  grad: 0.1008 (0.0955)  loss: 0.8073 (0.8124)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [37]  [2300/6250]  eta: 0:18:30  lr: 0.000092  grad: 0.0922 (0.0955)  loss: 0.8115 (0.8121)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [37]  [2400/6250]  eta: 0:18:00  lr: 0.000092  grad: 0.0910 (0.0958)  loss: 0.7988 (0.8119)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [37]  [2500/6250]  eta: 0:17:30  lr: 0.000092  grad: 0.0971 (0.0960)  loss: 0.8062 (0.8116)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [37]  [2600/6250]  eta: 0:17:01  lr: 0.000092  grad: 0.0970 (0.0964)  loss: 0.8048 (0.8114)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [37]  [2700/6250]  eta: 0:16:31  lr: 0.000092  grad: 0.0966 (0.0966)  loss: 0.8051 (0.8110)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [37]  [2800/6250]  eta: 0:16:02  lr: 0.000092  grad: 0.1069 (0.0969)  loss: 0.8012 (0.8107)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [37]  [2900/6250]  eta: 0:15:33  lr: 0.000092  grad: 0.1005 (0.0971)  loss: 0.8037 (0.8104)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [37]  [3000/6250]  eta: 0:15:04  lr: 0.000092  grad: 0.1004 (0.0972)  loss: 0.8010 (0.8102)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [37]  [3100/6250]  eta: 0:14:35  lr: 0.000092  grad: 0.0969 (0.0973)  loss: 0.8069 (0.8100)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [37]  [3200/6250]  eta: 0:14:07  lr: 0.000092  grad: 0.0959 (0.0975)  loss: 0.7987 (0.8098)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [37]  [3300/6250]  eta: 0:13:38  lr: 0.000092  grad: 0.0964 (0.0975)  loss: 0.8038 (0.8096)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [37]  [3400/6250]  eta: 0:13:10  lr: 0.000092  grad: 0.0945 (0.0976)  loss: 0.8074 (0.8094)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [37]  [3500/6250]  eta: 0:12:43  lr: 0.000092  grad: 0.0934 (0.0976)  loss: 0.8087 (0.8094)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [37]  [3600/6250]  eta: 0:12:17  lr: 0.000092  grad: 0.0995 (0.0977)  loss: 0.8034 (0.8092)  time: 0.4632  data: 0.1918  max mem: 26157
Train: [37]  [3700/6250]  eta: 0:11:49  lr: 0.000092  grad: 0.0970 (0.0979)  loss: 0.8048 (0.8090)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [37]  [3800/6250]  eta: 0:11:20  lr: 0.000092  grad: 0.1055 (0.0980)  loss: 0.8069 (0.8089)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [37]  [3900/6250]  eta: 0:10:57  lr: 0.000092  grad: 0.1040 (0.0981)  loss: 0.8070 (0.8088)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [37]  [4000/6250]  eta: 0:10:28  lr: 0.000092  grad: 0.0973 (0.0983)  loss: 0.8020 (0.8087)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [37]  [4100/6250]  eta: 0:10:02  lr: 0.000092  grad: 0.0961 (0.0984)  loss: 0.8079 (0.8085)  time: 0.4771  data: 0.2030  max mem: 26157
Train: [37]  [4200/6250]  eta: 0:09:33  lr: 0.000092  grad: 0.0985 (0.0985)  loss: 0.8097 (0.8084)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [37]  [4300/6250]  eta: 0:09:05  lr: 0.000092  grad: 0.0979 (0.0984)  loss: 0.8093 (0.8084)  time: 0.2962  data: 0.0117  max mem: 26157
Train: [37]  [4400/6250]  eta: 0:08:38  lr: 0.000092  grad: 0.0982 (0.0984)  loss: 0.8099 (0.8084)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [37]  [4500/6250]  eta: 0:08:10  lr: 0.000092  grad: 0.0893 (0.0983)  loss: 0.8037 (0.8084)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [37]  [4600/6250]  eta: 0:07:41  lr: 0.000092  grad: 0.0925 (0.0983)  loss: 0.8055 (0.8084)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [37]  [4700/6250]  eta: 0:07:13  lr: 0.000092  grad: 0.0963 (0.0983)  loss: 0.8084 (0.8083)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [37]  [4800/6250]  eta: 0:06:45  lr: 0.000092  grad: 0.0921 (0.0982)  loss: 0.8122 (0.8084)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [37]  [4900/6250]  eta: 0:06:17  lr: 0.000092  grad: 0.0936 (0.0982)  loss: 0.8170 (0.8085)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [37]  [5000/6250]  eta: 0:05:48  lr: 0.000092  grad: 0.0946 (0.0981)  loss: 0.8157 (0.8085)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [37]  [5100/6250]  eta: 0:05:20  lr: 0.000092  grad: 0.0912 (0.0980)  loss: 0.8136 (0.8086)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [37]  [5200/6250]  eta: 0:04:52  lr: 0.000092  grad: 0.0915 (0.0980)  loss: 0.8186 (0.8086)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [37]  [5300/6250]  eta: 0:04:24  lr: 0.000092  grad: 0.0949 (0.0980)  loss: 0.8143 (0.8086)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [37]  [5400/6250]  eta: 0:03:56  lr: 0.000092  grad: 0.0924 (0.0979)  loss: 0.8077 (0.8087)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [37]  [5500/6250]  eta: 0:03:28  lr: 0.000092  grad: 0.0934 (0.0979)  loss: 0.8125 (0.8087)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [37]  [5600/6250]  eta: 0:03:00  lr: 0.000092  grad: 0.0932 (0.0979)  loss: 0.8149 (0.8087)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [37]  [5700/6250]  eta: 0:02:32  lr: 0.000091  grad: 0.0960 (0.0979)  loss: 0.8149 (0.8088)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [37]  [5800/6250]  eta: 0:02:04  lr: 0.000091  grad: 0.1034 (0.0980)  loss: 0.8131 (0.8088)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [37]  [5900/6250]  eta: 0:01:37  lr: 0.000091  grad: 0.0992 (0.0981)  loss: 0.8051 (0.8088)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [37]  [6000/6250]  eta: 0:01:09  lr: 0.000091  grad: 0.0992 (0.0982)  loss: 0.8110 (0.8087)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [37]  [6100/6250]  eta: 0:00:41  lr: 0.000091  grad: 0.0977 (0.0982)  loss: 0.8079 (0.8087)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [37]  [6200/6250]  eta: 0:00:13  lr: 0.000091  grad: 0.0940 (0.0982)  loss: 0.8163 (0.8087)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [37]  [6249/6250]  eta: 0:00:00  lr: 0.000091  grad: 0.0978 (0.0983)  loss: 0.8036 (0.8087)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [37] Total time: 0:28:57 (0.2781 s / it)
Averaged stats: lr: 0.000091  grad: 0.0978 (0.0983)  loss: 0.8036 (0.8087)
Eval (hcp-train-subset): [37]  [ 0/62]  eta: 0:04:41  loss: 0.8373 (0.8373)  time: 4.5386  data: 4.4546  max mem: 26157
Eval (hcp-train-subset): [37]  [61/62]  eta: 0:00:00  loss: 0.8289 (0.8285)  time: 0.1185  data: 0.0336  max mem: 26157
Eval (hcp-train-subset): [37] Total time: 0:00:12 (0.2050 s / it)
Averaged stats (hcp-train-subset): loss: 0.8289 (0.8285)
Making plots (hcp-train-subset): example=59
Eval (hcp-val): [37]  [ 0/62]  eta: 0:03:45  loss: 0.8237 (0.8237)  time: 3.6301  data: 3.4898  max mem: 26157
Eval (hcp-val): [37]  [61/62]  eta: 0:00:00  loss: 0.8243 (0.8273)  time: 0.1096  data: 0.0270  max mem: 26157
Eval (hcp-val): [37] Total time: 0:00:12 (0.2057 s / it)
Averaged stats (hcp-val): loss: 0.8243 (0.8273)
Making plots (hcp-val): example=29
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [38]  [   0/6250]  eta: 7:06:26  lr: 0.000091  grad: 0.0593 (0.0593)  loss: 0.8689 (0.8689)  time: 4.0938  data: 3.7435  max mem: 26157
Train: [38]  [ 100/6250]  eta: 0:33:19  lr: 0.000091  grad: 0.1059 (0.1139)  loss: 0.8010 (0.8142)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [38]  [ 200/6250]  eta: 0:30:05  lr: 0.000091  grad: 0.1029 (0.1070)  loss: 0.8146 (0.8127)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [38]  [ 300/6250]  eta: 0:29:29  lr: 0.000091  grad: 0.0988 (0.1063)  loss: 0.8120 (0.8115)  time: 0.3695  data: 0.0814  max mem: 26157
Train: [38]  [ 400/6250]  eta: 0:28:34  lr: 0.000091  grad: 0.0983 (0.1049)  loss: 0.8107 (0.8117)  time: 0.3139  data: 0.0400  max mem: 26157
Train: [38]  [ 500/6250]  eta: 0:28:06  lr: 0.000091  grad: 0.0921 (0.1036)  loss: 0.8163 (0.8126)  time: 0.3818  data: 0.0986  max mem: 26157
Train: [38]  [ 600/6250]  eta: 0:27:15  lr: 0.000091  grad: 0.0909 (0.1026)  loss: 0.8221 (0.8129)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [38]  [ 700/6250]  eta: 0:26:52  lr: 0.000091  grad: 0.0884 (0.1010)  loss: 0.8106 (0.8129)  time: 0.3929  data: 0.1183  max mem: 26157
Train: [38]  [ 800/6250]  eta: 0:26:28  lr: 0.000091  grad: 0.0877 (0.1004)  loss: 0.8172 (0.8128)  time: 0.3946  data: 0.1068  max mem: 26157
Train: [38]  [ 900/6250]  eta: 0:26:00  lr: 0.000091  grad: 0.0910 (0.0997)  loss: 0.8127 (0.8127)  time: 0.2734  data: 0.0003  max mem: 26157
Train: [38]  [1000/6250]  eta: 0:25:21  lr: 0.000091  grad: 0.0910 (0.0992)  loss: 0.8149 (0.8127)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [38]  [1100/6250]  eta: 0:25:07  lr: 0.000091  grad: 0.0974 (0.0986)  loss: 0.8001 (0.8124)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [38]  [1200/6250]  eta: 0:24:29  lr: 0.000091  grad: 0.0968 (0.0983)  loss: 0.8092 (0.8124)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [38]  [1300/6250]  eta: 0:24:06  lr: 0.000091  grad: 0.0992 (0.0982)  loss: 0.8104 (0.8123)  time: 0.2697  data: 0.0001  max mem: 26157
Train: [38]  [1400/6250]  eta: 0:23:30  lr: 0.000091  grad: 0.0931 (0.0978)  loss: 0.8148 (0.8123)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [38]  [1500/6250]  eta: 0:22:54  lr: 0.000091  grad: 0.0932 (0.0978)  loss: 0.8109 (0.8122)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [38]  [1600/6250]  eta: 0:22:19  lr: 0.000091  grad: 0.0954 (0.0978)  loss: 0.8121 (0.8120)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [38]  [1700/6250]  eta: 0:21:46  lr: 0.000091  grad: 0.0950 (0.0977)  loss: 0.8043 (0.8118)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [38]  [1800/6250]  eta: 0:21:13  lr: 0.000091  grad: 0.0955 (0.0977)  loss: 0.8095 (0.8117)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [38]  [1900/6250]  eta: 0:20:41  lr: 0.000091  grad: 0.0974 (0.0977)  loss: 0.8061 (0.8116)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [38]  [2000/6250]  eta: 0:20:09  lr: 0.000091  grad: 0.1018 (0.0978)  loss: 0.8073 (0.8114)  time: 0.2781  data: 0.0002  max mem: 26157
Train: [38]  [2100/6250]  eta: 0:19:38  lr: 0.000091  grad: 0.0967 (0.0980)  loss: 0.8065 (0.8113)  time: 0.2707  data: 0.0003  max mem: 26157
Train: [38]  [2200/6250]  eta: 0:19:07  lr: 0.000091  grad: 0.0955 (0.0980)  loss: 0.8093 (0.8113)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [38]  [2300/6250]  eta: 0:18:41  lr: 0.000091  grad: 0.0981 (0.0981)  loss: 0.8057 (0.8111)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [38]  [2400/6250]  eta: 0:18:10  lr: 0.000091  grad: 0.0962 (0.0982)  loss: 0.8101 (0.8110)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [38]  [2500/6250]  eta: 0:17:39  lr: 0.000091  grad: 0.0992 (0.0983)  loss: 0.8043 (0.8109)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [38]  [2600/6250]  eta: 0:17:09  lr: 0.000091  grad: 0.0977 (0.0983)  loss: 0.8083 (0.8109)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [38]  [2700/6250]  eta: 0:16:40  lr: 0.000091  grad: 0.0947 (0.0984)  loss: 0.8056 (0.8107)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [38]  [2800/6250]  eta: 0:16:11  lr: 0.000091  grad: 0.0924 (0.0984)  loss: 0.8030 (0.8106)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [38]  [2900/6250]  eta: 0:15:42  lr: 0.000090  grad: 0.1003 (0.0984)  loss: 0.8024 (0.8104)  time: 0.2755  data: 0.0002  max mem: 26157
Train: [38]  [3000/6250]  eta: 0:15:13  lr: 0.000090  grad: 0.0980 (0.0985)  loss: 0.8065 (0.8103)  time: 0.2693  data: 0.0001  max mem: 26157
Train: [38]  [3100/6250]  eta: 0:14:44  lr: 0.000090  grad: 0.0948 (0.0987)  loss: 0.8052 (0.8102)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [38]  [3200/6250]  eta: 0:14:14  lr: 0.000090  grad: 0.0989 (0.0987)  loss: 0.8043 (0.8100)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [38]  [3300/6250]  eta: 0:13:45  lr: 0.000090  grad: 0.1060 (0.0988)  loss: 0.7974 (0.8098)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [38]  [3400/6250]  eta: 0:13:16  lr: 0.000090  grad: 0.0957 (0.0988)  loss: 0.8033 (0.8098)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [38]  [3500/6250]  eta: 0:12:48  lr: 0.000090  grad: 0.1002 (0.0990)  loss: 0.8043 (0.8096)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [38]  [3600/6250]  eta: 0:12:19  lr: 0.000090  grad: 0.1017 (0.0991)  loss: 0.7972 (0.8095)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [38]  [3700/6250]  eta: 0:11:50  lr: 0.000090  grad: 0.1010 (0.0991)  loss: 0.8028 (0.8094)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [38]  [3800/6250]  eta: 0:11:22  lr: 0.000090  grad: 0.1022 (0.0992)  loss: 0.8008 (0.8092)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [38]  [3900/6250]  eta: 0:10:54  lr: 0.000090  grad: 0.0950 (0.0992)  loss: 0.7997 (0.8091)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [38]  [4000/6250]  eta: 0:10:25  lr: 0.000090  grad: 0.1020 (0.0993)  loss: 0.8005 (0.8090)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [38]  [4100/6250]  eta: 0:09:57  lr: 0.000090  grad: 0.0974 (0.0993)  loss: 0.8054 (0.8088)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [38]  [4200/6250]  eta: 0:09:29  lr: 0.000090  grad: 0.0938 (0.0993)  loss: 0.8079 (0.8087)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [38]  [4300/6250]  eta: 0:09:01  lr: 0.000090  grad: 0.0962 (0.0993)  loss: 0.8054 (0.8086)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [38]  [4400/6250]  eta: 0:08:33  lr: 0.000090  grad: 0.0964 (0.0994)  loss: 0.8054 (0.8086)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [38]  [4500/6250]  eta: 0:08:05  lr: 0.000090  grad: 0.1016 (0.0996)  loss: 0.8038 (0.8085)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [38]  [4600/6250]  eta: 0:07:37  lr: 0.000090  grad: 0.0977 (0.0996)  loss: 0.8082 (0.8085)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [38]  [4700/6250]  eta: 0:07:09  lr: 0.000090  grad: 0.0942 (0.0996)  loss: 0.8120 (0.8085)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [38]  [4800/6250]  eta: 0:06:41  lr: 0.000090  grad: 0.0909 (0.0997)  loss: 0.8135 (0.8085)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [38]  [4900/6250]  eta: 0:06:13  lr: 0.000090  grad: 0.1014 (0.0997)  loss: 0.8150 (0.8085)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [38]  [5000/6250]  eta: 0:05:45  lr: 0.000090  grad: 0.0918 (0.0997)  loss: 0.8100 (0.8085)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [38]  [5100/6250]  eta: 0:05:17  lr: 0.000090  grad: 0.1018 (0.0997)  loss: 0.8058 (0.8085)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [38]  [5200/6250]  eta: 0:04:49  lr: 0.000090  grad: 0.0956 (0.0997)  loss: 0.8107 (0.8085)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [38]  [5300/6250]  eta: 0:04:22  lr: 0.000090  grad: 0.0968 (0.0997)  loss: 0.8097 (0.8085)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [38]  [5400/6250]  eta: 0:03:54  lr: 0.000090  grad: 0.0921 (0.0996)  loss: 0.8095 (0.8085)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [38]  [5500/6250]  eta: 0:03:26  lr: 0.000090  grad: 0.0986 (0.0997)  loss: 0.8036 (0.8085)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [38]  [5600/6250]  eta: 0:02:59  lr: 0.000090  grad: 0.1005 (0.0998)  loss: 0.8086 (0.8084)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [38]  [5700/6250]  eta: 0:02:31  lr: 0.000090  grad: 0.1104 (0.0998)  loss: 0.7977 (0.8083)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [38]  [5800/6250]  eta: 0:02:03  lr: 0.000090  grad: 0.0988 (0.0999)  loss: 0.8044 (0.8083)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [38]  [5900/6250]  eta: 0:01:36  lr: 0.000090  grad: 0.0926 (0.1000)  loss: 0.8100 (0.8081)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [38]  [6000/6250]  eta: 0:01:08  lr: 0.000090  grad: 0.0977 (0.1001)  loss: 0.8054 (0.8080)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [38]  [6100/6250]  eta: 0:00:41  lr: 0.000090  grad: 0.1071 (0.1002)  loss: 0.8011 (0.8080)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [38]  [6200/6250]  eta: 0:00:13  lr: 0.000089  grad: 0.1031 (0.1003)  loss: 0.8002 (0.8078)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [38]  [6249/6250]  eta: 0:00:00  lr: 0.000089  grad: 0.0980 (0.1003)  loss: 0.8066 (0.8078)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [38] Total time: 0:28:44 (0.2760 s / it)
Averaged stats: lr: 0.000089  grad: 0.0980 (0.1003)  loss: 0.8066 (0.8078)
Eval (hcp-train-subset): [38]  [ 0/62]  eta: 0:03:19  loss: 0.8383 (0.8383)  time: 3.2142  data: 3.0904  max mem: 26157
Eval (hcp-train-subset): [38]  [61/62]  eta: 0:00:00  loss: 0.8278 (0.8294)  time: 0.1384  data: 0.0554  max mem: 26157
Eval (hcp-train-subset): [38] Total time: 0:00:13 (0.2167 s / it)
Averaged stats (hcp-train-subset): loss: 0.8278 (0.8294)
Making plots (hcp-train-subset): example=35
Eval (hcp-val): [38]  [ 0/62]  eta: 0:05:27  loss: 0.8241 (0.8241)  time: 5.2855  data: 5.1949  max mem: 26157
Eval (hcp-val): [38]  [61/62]  eta: 0:00:00  loss: 0.8274 (0.8275)  time: 0.1305  data: 0.0453  max mem: 26157
Eval (hcp-val): [38] Total time: 0:00:14 (0.2261 s / it)
Averaged stats (hcp-val): loss: 0.8274 (0.8275)
Making plots (hcp-val): example=62
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [39]  [   0/6250]  eta: 8:08:53  lr: 0.000089  grad: 0.0687 (0.0687)  loss: 0.8752 (0.8752)  time: 4.6933  data: 4.3485  max mem: 26157
Train: [39]  [ 100/6250]  eta: 0:33:28  lr: 0.000089  grad: 0.0999 (0.1212)  loss: 0.8222 (0.8229)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [39]  [ 200/6250]  eta: 0:30:08  lr: 0.000089  grad: 0.0962 (0.1112)  loss: 0.8205 (0.8200)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [39]  [ 300/6250]  eta: 0:28:45  lr: 0.000089  grad: 0.1015 (0.1078)  loss: 0.8103 (0.8196)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [39]  [ 400/6250]  eta: 0:27:48  lr: 0.000089  grad: 0.0949 (0.1059)  loss: 0.8151 (0.8183)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [39]  [ 500/6250]  eta: 0:27:02  lr: 0.000089  grad: 0.0971 (0.1046)  loss: 0.8157 (0.8168)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [39]  [ 600/6250]  eta: 0:26:21  lr: 0.000089  grad: 0.0937 (0.1035)  loss: 0.8137 (0.8161)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [39]  [ 700/6250]  eta: 0:25:46  lr: 0.000089  grad: 0.0980 (0.1025)  loss: 0.8126 (0.8158)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [39]  [ 800/6250]  eta: 0:25:12  lr: 0.000089  grad: 0.0950 (0.1024)  loss: 0.8083 (0.8149)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [39]  [ 900/6250]  eta: 0:24:40  lr: 0.000089  grad: 0.0941 (0.1025)  loss: 0.8170 (0.8146)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [39]  [1000/6250]  eta: 0:24:09  lr: 0.000089  grad: 0.0925 (0.1018)  loss: 0.8097 (0.8143)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [39]  [1100/6250]  eta: 0:23:39  lr: 0.000089  grad: 0.0933 (0.1012)  loss: 0.8160 (0.8140)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [39]  [1200/6250]  eta: 0:23:09  lr: 0.000089  grad: 0.0872 (0.1009)  loss: 0.8139 (0.8135)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [39]  [1300/6250]  eta: 0:22:40  lr: 0.000089  grad: 0.1045 (0.1009)  loss: 0.8062 (0.8130)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [39]  [1400/6250]  eta: 0:22:11  lr: 0.000089  grad: 0.0966 (0.1011)  loss: 0.8067 (0.8125)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [39]  [1500/6250]  eta: 0:21:45  lr: 0.000089  grad: 0.0925 (0.1012)  loss: 0.8058 (0.8123)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [39]  [1600/6250]  eta: 0:21:16  lr: 0.000089  grad: 0.0968 (0.1012)  loss: 0.8049 (0.8117)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [39]  [1700/6250]  eta: 0:21:02  lr: 0.000089  grad: 0.0932 (0.1013)  loss: 0.8054 (0.8112)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [39]  [1800/6250]  eta: 0:20:32  lr: 0.000089  grad: 0.0992 (0.1014)  loss: 0.8011 (0.8107)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [39]  [1900/6250]  eta: 0:20:09  lr: 0.000089  grad: 0.1078 (0.1015)  loss: 0.8046 (0.8104)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [39]  [2000/6250]  eta: 0:19:39  lr: 0.000089  grad: 0.1019 (0.1017)  loss: 0.8029 (0.8100)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [39]  [2100/6250]  eta: 0:19:10  lr: 0.000089  grad: 0.1044 (0.1020)  loss: 0.8034 (0.8095)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [39]  [2200/6250]  eta: 0:18:41  lr: 0.000089  grad: 0.0962 (0.1021)  loss: 0.8006 (0.8090)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [39]  [2300/6250]  eta: 0:18:12  lr: 0.000089  grad: 0.1042 (0.1022)  loss: 0.8000 (0.8086)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [39]  [2400/6250]  eta: 0:17:43  lr: 0.000089  grad: 0.1041 (0.1024)  loss: 0.8059 (0.8083)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [39]  [2500/6250]  eta: 0:17:20  lr: 0.000089  grad: 0.0970 (0.1025)  loss: 0.8048 (0.8081)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [39]  [2600/6250]  eta: 0:16:51  lr: 0.000089  grad: 0.1115 (0.1025)  loss: 0.8003 (0.8078)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [39]  [2700/6250]  eta: 0:16:22  lr: 0.000089  grad: 0.1010 (0.1027)  loss: 0.7994 (0.8076)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [39]  [2800/6250]  eta: 0:15:54  lr: 0.000089  grad: 0.1046 (0.1028)  loss: 0.8080 (0.8074)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [39]  [2900/6250]  eta: 0:15:25  lr: 0.000089  grad: 0.0955 (0.1028)  loss: 0.8110 (0.8073)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [39]  [3000/6250]  eta: 0:14:57  lr: 0.000089  grad: 0.1055 (0.1028)  loss: 0.7971 (0.8072)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [39]  [3100/6250]  eta: 0:14:29  lr: 0.000089  grad: 0.1031 (0.1031)  loss: 0.8032 (0.8072)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [39]  [3200/6250]  eta: 0:14:01  lr: 0.000089  grad: 0.1000 (0.1032)  loss: 0.8026 (0.8071)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [39]  [3300/6250]  eta: 0:13:32  lr: 0.000088  grad: 0.1050 (0.1033)  loss: 0.8048 (0.8070)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [39]  [3400/6250]  eta: 0:13:04  lr: 0.000088  grad: 0.1005 (0.1034)  loss: 0.8068 (0.8070)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [39]  [3500/6250]  eta: 0:12:36  lr: 0.000088  grad: 0.1045 (0.1035)  loss: 0.8017 (0.8069)  time: 0.2744  data: 0.0003  max mem: 26157
Train: [39]  [3600/6250]  eta: 0:12:09  lr: 0.000088  grad: 0.0935 (0.1034)  loss: 0.8040 (0.8070)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [39]  [3700/6250]  eta: 0:11:41  lr: 0.000088  grad: 0.0929 (0.1033)  loss: 0.8056 (0.8070)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [39]  [3800/6250]  eta: 0:11:13  lr: 0.000088  grad: 0.1034 (0.1034)  loss: 0.8033 (0.8069)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [39]  [3900/6250]  eta: 0:10:45  lr: 0.000088  grad: 0.1014 (0.1034)  loss: 0.8080 (0.8070)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [39]  [4000/6250]  eta: 0:10:20  lr: 0.000088  grad: 0.1016 (0.1035)  loss: 0.8044 (0.8070)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [39]  [4100/6250]  eta: 0:09:52  lr: 0.000088  grad: 0.1026 (0.1035)  loss: 0.8096 (0.8070)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [39]  [4200/6250]  eta: 0:09:28  lr: 0.000088  grad: 0.0986 (0.1035)  loss: 0.8070 (0.8070)  time: 0.6547  data: 0.3766  max mem: 26157
Train: [39]  [4300/6250]  eta: 0:09:00  lr: 0.000088  grad: 0.0999 (0.1035)  loss: 0.8055 (0.8070)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [39]  [4400/6250]  eta: 0:08:32  lr: 0.000088  grad: 0.1031 (0.1035)  loss: 0.8090 (0.8070)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [39]  [4500/6250]  eta: 0:08:04  lr: 0.000088  grad: 0.0998 (0.1035)  loss: 0.8066 (0.8071)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [39]  [4600/6250]  eta: 0:07:36  lr: 0.000088  grad: 0.1026 (0.1035)  loss: 0.8038 (0.8071)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [39]  [4700/6250]  eta: 0:07:08  lr: 0.000088  grad: 0.0991 (0.1036)  loss: 0.8038 (0.8071)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [39]  [4800/6250]  eta: 0:06:40  lr: 0.000088  grad: 0.1033 (0.1036)  loss: 0.8135 (0.8072)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [39]  [4900/6250]  eta: 0:06:12  lr: 0.000088  grad: 0.0989 (0.1037)  loss: 0.8162 (0.8072)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [39]  [5000/6250]  eta: 0:05:45  lr: 0.000088  grad: 0.0958 (0.1036)  loss: 0.8141 (0.8073)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [39]  [5100/6250]  eta: 0:05:17  lr: 0.000088  grad: 0.1045 (0.1036)  loss: 0.8086 (0.8074)  time: 0.3019  data: 0.0302  max mem: 26157
Train: [39]  [5200/6250]  eta: 0:04:49  lr: 0.000088  grad: 0.1006 (0.1035)  loss: 0.8156 (0.8075)  time: 0.2722  data: 0.0003  max mem: 26157
Train: [39]  [5300/6250]  eta: 0:04:23  lr: 0.000088  grad: 0.0960 (0.1035)  loss: 0.8203 (0.8076)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [39]  [5400/6250]  eta: 0:03:55  lr: 0.000088  grad: 0.0974 (0.1034)  loss: 0.8081 (0.8076)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [39]  [5500/6250]  eta: 0:03:27  lr: 0.000088  grad: 0.0993 (0.1035)  loss: 0.8101 (0.8077)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [39]  [5600/6250]  eta: 0:02:59  lr: 0.000088  grad: 0.1047 (0.1035)  loss: 0.8123 (0.8078)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [39]  [5700/6250]  eta: 0:02:31  lr: 0.000088  grad: 0.1012 (0.1035)  loss: 0.8118 (0.8078)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [39]  [5800/6250]  eta: 0:02:04  lr: 0.000088  grad: 0.0941 (0.1034)  loss: 0.8077 (0.8078)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [39]  [5900/6250]  eta: 0:01:36  lr: 0.000088  grad: 0.0970 (0.1034)  loss: 0.8064 (0.8079)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [39]  [6000/6250]  eta: 0:01:08  lr: 0.000088  grad: 0.0946 (0.1034)  loss: 0.8135 (0.8079)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [39]  [6100/6250]  eta: 0:00:41  lr: 0.000088  grad: 0.0967 (0.1034)  loss: 0.8073 (0.8079)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [39]  [6200/6250]  eta: 0:00:13  lr: 0.000088  grad: 0.0964 (0.1034)  loss: 0.8145 (0.8080)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [39]  [6249/6250]  eta: 0:00:00  lr: 0.000088  grad: 0.0982 (0.1034)  loss: 0.8093 (0.8080)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [39] Total time: 0:28:49 (0.2766 s / it)
Averaged stats: lr: 0.000088  grad: 0.0982 (0.1034)  loss: 0.8093 (0.8080)
Eval (hcp-train-subset): [39]  [ 0/62]  eta: 0:04:06  loss: 0.8406 (0.8406)  time: 3.9739  data: 3.8735  max mem: 26157
Eval (hcp-train-subset): [39]  [61/62]  eta: 0:00:00  loss: 0.8278 (0.8295)  time: 0.1413  data: 0.0564  max mem: 26157
Eval (hcp-train-subset): [39] Total time: 0:00:12 (0.2092 s / it)
Averaged stats (hcp-train-subset): loss: 0.8278 (0.8295)
Making plots (hcp-train-subset): example=47
Eval (hcp-val): [39]  [ 0/62]  eta: 0:05:41  loss: 0.8248 (0.8248)  time: 5.5078  data: 5.4230  max mem: 26157
Eval (hcp-val): [39]  [61/62]  eta: 0:00:00  loss: 0.8254 (0.8272)  time: 0.1181  data: 0.0350  max mem: 26157
Eval (hcp-val): [39] Total time: 0:00:12 (0.2024 s / it)
Averaged stats (hcp-val): loss: 0.8254 (0.8272)
Making plots (hcp-val): example=36
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [40]  [   0/6250]  eta: 10:48:10  lr: 0.000088  grad: 0.1008 (0.1008)  loss: 0.8240 (0.8240)  time: 6.2224  data: 5.9428  max mem: 26157
Train: [40]  [ 100/6250]  eta: 0:34:24  lr: 0.000088  grad: 0.1096 (0.1158)  loss: 0.8047 (0.8133)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [40]  [ 200/6250]  eta: 0:30:38  lr: 0.000088  grad: 0.1047 (0.1113)  loss: 0.8120 (0.8090)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [40]  [ 300/6250]  eta: 0:29:03  lr: 0.000088  grad: 0.1049 (0.1094)  loss: 0.8010 (0.8067)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [40]  [ 400/6250]  eta: 0:28:02  lr: 0.000087  grad: 0.0955 (0.1069)  loss: 0.8045 (0.8062)  time: 0.2679  data: 0.0003  max mem: 26157
Train: [40]  [ 500/6250]  eta: 0:27:14  lr: 0.000087  grad: 0.0970 (0.1054)  loss: 0.8106 (0.8067)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [40]  [ 600/6250]  eta: 0:26:32  lr: 0.000087  grad: 0.0980 (0.1048)  loss: 0.8097 (0.8071)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [40]  [ 700/6250]  eta: 0:25:55  lr: 0.000087  grad: 0.0927 (0.1045)  loss: 0.8212 (0.8077)  time: 0.2722  data: 0.0003  max mem: 26157
Train: [40]  [ 800/6250]  eta: 0:25:22  lr: 0.000087  grad: 0.1007 (0.1042)  loss: 0.8085 (0.8076)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [40]  [ 900/6250]  eta: 0:24:48  lr: 0.000087  grad: 0.1023 (0.1046)  loss: 0.8028 (0.8072)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [40]  [1000/6250]  eta: 0:24:17  lr: 0.000087  grad: 0.1001 (0.1047)  loss: 0.8098 (0.8070)  time: 0.2738  data: 0.0002  max mem: 26157
Train: [40]  [1100/6250]  eta: 0:23:49  lr: 0.000087  grad: 0.0995 (0.1047)  loss: 0.8094 (0.8068)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [40]  [1200/6250]  eta: 0:23:18  lr: 0.000087  grad: 0.0988 (0.1045)  loss: 0.8072 (0.8066)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [40]  [1300/6250]  eta: 0:22:48  lr: 0.000087  grad: 0.1035 (0.1045)  loss: 0.8029 (0.8063)  time: 0.2698  data: 0.0003  max mem: 26157
Train: [40]  [1400/6250]  eta: 0:22:30  lr: 0.000087  grad: 0.1002 (0.1043)  loss: 0.7975 (0.8061)  time: 0.4139  data: 0.1395  max mem: 26157
Train: [40]  [1500/6250]  eta: 0:21:59  lr: 0.000087  grad: 0.0953 (0.1042)  loss: 0.8063 (0.8059)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [40]  [1600/6250]  eta: 0:21:30  lr: 0.000087  grad: 0.0950 (0.1041)  loss: 0.8085 (0.8059)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [40]  [1700/6250]  eta: 0:21:07  lr: 0.000087  grad: 0.1037 (0.1039)  loss: 0.7993 (0.8057)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [40]  [1800/6250]  eta: 0:20:37  lr: 0.000087  grad: 0.0994 (0.1040)  loss: 0.7936 (0.8054)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [40]  [1900/6250]  eta: 0:20:22  lr: 0.000087  grad: 0.1056 (0.1039)  loss: 0.7999 (0.8052)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [40]  [2000/6250]  eta: 0:19:52  lr: 0.000087  grad: 0.1040 (0.1038)  loss: 0.8057 (0.8051)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [40]  [2100/6250]  eta: 0:19:22  lr: 0.000087  grad: 0.0989 (0.1037)  loss: 0.8034 (0.8050)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [40]  [2200/6250]  eta: 0:18:52  lr: 0.000087  grad: 0.1032 (0.1038)  loss: 0.7971 (0.8048)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [40]  [2300/6250]  eta: 0:18:22  lr: 0.000087  grad: 0.1023 (0.1038)  loss: 0.7956 (0.8047)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [40]  [2400/6250]  eta: 0:17:53  lr: 0.000087  grad: 0.1038 (0.1039)  loss: 0.8004 (0.8045)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [40]  [2500/6250]  eta: 0:17:24  lr: 0.000087  grad: 0.1040 (0.1041)  loss: 0.7984 (0.8043)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [40]  [2600/6250]  eta: 0:16:55  lr: 0.000087  grad: 0.1051 (0.1045)  loss: 0.7952 (0.8041)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [40]  [2700/6250]  eta: 0:16:26  lr: 0.000087  grad: 0.1101 (0.1045)  loss: 0.8012 (0.8039)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [40]  [2800/6250]  eta: 0:15:57  lr: 0.000087  grad: 0.1012 (0.1047)  loss: 0.8039 (0.8038)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [40]  [2900/6250]  eta: 0:15:29  lr: 0.000087  grad: 0.1011 (0.1049)  loss: 0.8025 (0.8037)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [40]  [3000/6250]  eta: 0:15:00  lr: 0.000087  grad: 0.1031 (0.1051)  loss: 0.7974 (0.8035)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [40]  [3100/6250]  eta: 0:14:32  lr: 0.000087  grad: 0.1005 (0.1051)  loss: 0.8013 (0.8034)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [40]  [3200/6250]  eta: 0:14:05  lr: 0.000087  grad: 0.1061 (0.1052)  loss: 0.7945 (0.8033)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [40]  [3300/6250]  eta: 0:13:38  lr: 0.000087  grad: 0.1105 (0.1052)  loss: 0.8034 (0.8032)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [40]  [3400/6250]  eta: 0:13:10  lr: 0.000087  grad: 0.1033 (0.1052)  loss: 0.8112 (0.8032)  time: 0.2759  data: 0.0002  max mem: 26157
Train: [40]  [3500/6250]  eta: 0:12:47  lr: 0.000087  grad: 0.1064 (0.1052)  loss: 0.8006 (0.8032)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [40]  [3600/6250]  eta: 0:12:18  lr: 0.000087  grad: 0.0989 (0.1052)  loss: 0.8084 (0.8032)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [40]  [3700/6250]  eta: 0:11:51  lr: 0.000086  grad: 0.0987 (0.1052)  loss: 0.8012 (0.8032)  time: 0.3867  data: 0.1151  max mem: 26157
Train: [40]  [3800/6250]  eta: 0:11:23  lr: 0.000086  grad: 0.1075 (0.1051)  loss: 0.8064 (0.8033)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [40]  [3900/6250]  eta: 0:10:54  lr: 0.000086  grad: 0.1013 (0.1051)  loss: 0.8018 (0.8034)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [40]  [4000/6250]  eta: 0:10:28  lr: 0.000086  grad: 0.1008 (0.1050)  loss: 0.8085 (0.8035)  time: 0.4487  data: 0.1732  max mem: 26157
Train: [40]  [4100/6250]  eta: 0:09:59  lr: 0.000086  grad: 0.1048 (0.1050)  loss: 0.8071 (0.8036)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [40]  [4200/6250]  eta: 0:09:31  lr: 0.000086  grad: 0.1022 (0.1051)  loss: 0.8027 (0.8037)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [40]  [4300/6250]  eta: 0:09:03  lr: 0.000086  grad: 0.1098 (0.1051)  loss: 0.7996 (0.8037)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [40]  [4400/6250]  eta: 0:08:35  lr: 0.000086  grad: 0.1060 (0.1051)  loss: 0.7992 (0.8038)  time: 0.2755  data: 0.0002  max mem: 26157
Train: [40]  [4500/6250]  eta: 0:08:07  lr: 0.000086  grad: 0.1091 (0.1052)  loss: 0.8068 (0.8039)  time: 0.2730  data: 0.0003  max mem: 26157
Train: [40]  [4600/6250]  eta: 0:07:39  lr: 0.000086  grad: 0.1076 (0.1053)  loss: 0.8045 (0.8040)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [40]  [4700/6250]  eta: 0:07:11  lr: 0.000086  grad: 0.1036 (0.1052)  loss: 0.7998 (0.8041)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [40]  [4800/6250]  eta: 0:06:43  lr: 0.000086  grad: 0.1003 (0.1051)  loss: 0.8135 (0.8043)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [40]  [4900/6250]  eta: 0:06:15  lr: 0.000086  grad: 0.1045 (0.1051)  loss: 0.8120 (0.8043)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [40]  [5000/6250]  eta: 0:05:47  lr: 0.000086  grad: 0.1072 (0.1053)  loss: 0.8006 (0.8044)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [40]  [5100/6250]  eta: 0:05:19  lr: 0.000086  grad: 0.0932 (0.1053)  loss: 0.8105 (0.8044)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [40]  [5200/6250]  eta: 0:04:51  lr: 0.000086  grad: 0.1088 (0.1053)  loss: 0.8061 (0.8045)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [40]  [5300/6250]  eta: 0:04:23  lr: 0.000086  grad: 0.1046 (0.1052)  loss: 0.8025 (0.8045)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [40]  [5400/6250]  eta: 0:03:56  lr: 0.000086  grad: 0.1038 (0.1052)  loss: 0.8125 (0.8046)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [40]  [5500/6250]  eta: 0:03:29  lr: 0.000086  grad: 0.1013 (0.1051)  loss: 0.8043 (0.8047)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [40]  [5600/6250]  eta: 0:03:01  lr: 0.000086  grad: 0.0988 (0.1051)  loss: 0.8117 (0.8047)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [40]  [5700/6250]  eta: 0:02:33  lr: 0.000086  grad: 0.1063 (0.1050)  loss: 0.8047 (0.8048)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [40]  [5800/6250]  eta: 0:02:05  lr: 0.000086  grad: 0.1057 (0.1051)  loss: 0.8022 (0.8048)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [40]  [5900/6250]  eta: 0:01:37  lr: 0.000086  grad: 0.1010 (0.1050)  loss: 0.8147 (0.8048)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [40]  [6000/6250]  eta: 0:01:09  lr: 0.000086  grad: 0.1031 (0.1050)  loss: 0.8022 (0.8048)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [40]  [6100/6250]  eta: 0:00:41  lr: 0.000086  grad: 0.1028 (0.1050)  loss: 0.8083 (0.8049)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [40]  [6200/6250]  eta: 0:00:13  lr: 0.000086  grad: 0.0988 (0.1049)  loss: 0.8086 (0.8049)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [40]  [6249/6250]  eta: 0:00:00  lr: 0.000086  grad: 0.1027 (0.1049)  loss: 0.8044 (0.8049)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [40] Total time: 0:29:07 (0.2796 s / it)
Averaged stats: lr: 0.000086  grad: 0.1027 (0.1049)  loss: 0.8044 (0.8049)
Eval (hcp-train-subset): [40]  [ 0/62]  eta: 0:03:44  loss: 0.8433 (0.8433)  time: 3.6229  data: 3.5128  max mem: 26157
Eval (hcp-train-subset): [40]  [61/62]  eta: 0:00:00  loss: 0.8234 (0.8269)  time: 0.1132  data: 0.0303  max mem: 26157
Eval (hcp-train-subset): [40] Total time: 0:00:13 (0.2185 s / it)
Averaged stats (hcp-train-subset): loss: 0.8234 (0.8269)
Making plots (hcp-train-subset): example=33
Eval (hcp-val): [40]  [ 0/62]  eta: 0:05:00  loss: 0.8219 (0.8219)  time: 4.8496  data: 4.7502  max mem: 26157
Eval (hcp-val): [40]  [61/62]  eta: 0:00:00  loss: 0.8254 (0.8263)  time: 0.1302  data: 0.0456  max mem: 26157
Eval (hcp-val): [40] Total time: 0:00:13 (0.2235 s / it)
Averaged stats (hcp-val): loss: 0.8254 (0.8263)
Making plots (hcp-val): example=59
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [41]  [   0/6250]  eta: 12:15:30  lr: 0.000086  grad: nan (nan)  loss: 0.8215 (0.8215)  time: 7.0608  data: 6.7901  max mem: 26157
Train: [41]  [ 100/6250]  eta: 0:34:44  lr: 0.000086  grad: 0.0992 (0.1171)  loss: 0.8166 (0.8234)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [41]  [ 200/6250]  eta: 0:30:42  lr: 0.000086  grad: 0.0917 (0.1091)  loss: 0.8187 (0.8216)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [41]  [ 300/6250]  eta: 0:29:02  lr: 0.000086  grad: 0.0914 (0.1064)  loss: 0.8184 (0.8202)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [41]  [ 400/6250]  eta: 0:27:59  lr: 0.000086  grad: 0.1040 (0.1046)  loss: 0.8176 (0.8191)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [41]  [ 500/6250]  eta: 0:27:10  lr: 0.000086  grad: 0.0986 (0.1037)  loss: 0.8144 (0.8184)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [41]  [ 600/6250]  eta: 0:26:28  lr: 0.000086  grad: 0.0950 (0.1038)  loss: 0.8140 (0.8170)  time: 0.2692  data: 0.0001  max mem: 26157
Train: [41]  [ 700/6250]  eta: 0:25:50  lr: 0.000085  grad: 0.0924 (0.1034)  loss: 0.8095 (0.8161)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [41]  [ 800/6250]  eta: 0:25:16  lr: 0.000085  grad: 0.1028 (0.1030)  loss: 0.8083 (0.8152)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [41]  [ 900/6250]  eta: 0:24:43  lr: 0.000085  grad: 0.0975 (0.1027)  loss: 0.8120 (0.8146)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [41]  [1000/6250]  eta: 0:24:11  lr: 0.000085  grad: 0.1009 (0.1031)  loss: 0.8133 (0.8140)  time: 0.2692  data: 0.0003  max mem: 26157
Train: [41]  [1100/6250]  eta: 0:23:40  lr: 0.000085  grad: 0.1006 (0.1032)  loss: 0.8064 (0.8134)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [41]  [1200/6250]  eta: 0:23:14  lr: 0.000085  grad: 0.0974 (0.1032)  loss: 0.8114 (0.8128)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [41]  [1300/6250]  eta: 0:22:45  lr: 0.000085  grad: 0.0977 (0.1032)  loss: 0.8072 (0.8123)  time: 0.2728  data: 0.0003  max mem: 26157
Train: [41]  [1400/6250]  eta: 0:22:15  lr: 0.000085  grad: 0.1011 (0.1032)  loss: 0.7970 (0.8118)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [41]  [1500/6250]  eta: 0:22:23  lr: 0.000085  grad: 0.0999 (0.1037)  loss: 0.8084 (0.8114)  time: 0.4355  data: 0.1616  max mem: 26157
Train: [41]  [1600/6250]  eta: 0:21:51  lr: 0.000085  grad: 0.0972 (0.1037)  loss: 0.8110 (0.8112)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [41]  [1700/6250]  eta: 0:21:19  lr: 0.000085  grad: 0.1022 (0.1037)  loss: 0.8130 (0.8111)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [41]  [1800/6250]  eta: 0:20:48  lr: 0.000085  grad: 0.1042 (0.1037)  loss: 0.8016 (0.8109)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [41]  [1900/6250]  eta: 0:20:17  lr: 0.000085  grad: 0.1011 (0.1037)  loss: 0.8074 (0.8108)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [41]  [2000/6250]  eta: 0:19:47  lr: 0.000085  grad: 0.1080 (0.1038)  loss: 0.8110 (0.8107)  time: 0.2720  data: 0.0003  max mem: 26157
Train: [41]  [2100/6250]  eta: 0:19:22  lr: 0.000085  grad: 0.1068 (0.1037)  loss: 0.8049 (0.8106)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [41]  [2200/6250]  eta: 0:18:55  lr: 0.000085  grad: 0.1037 (0.1043)  loss: 0.8079 (0.8105)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [41]  [2300/6250]  eta: 0:18:27  lr: 0.000085  grad: 0.1021 (0.1045)  loss: 0.8097 (0.8103)  time: 0.2993  data: 0.0003  max mem: 26157
Train: [41]  [2400/6250]  eta: 0:18:04  lr: 0.000085  grad: 0.1084 (0.1047)  loss: 0.8029 (0.8101)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [41]  [2500/6250]  eta: 0:17:51  lr: 0.000085  grad: 0.1051 (0.1046)  loss: 0.8001 (0.8100)  time: 0.8263  data: 0.5538  max mem: 26157
Train: [41]  [2600/6250]  eta: 0:17:20  lr: 0.000085  grad: 0.1016 (0.1046)  loss: 0.8055 (0.8098)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [41]  [2700/6250]  eta: 0:16:53  lr: 0.000085  grad: 0.0992 (0.1047)  loss: 0.8051 (0.8096)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [41]  [2800/6250]  eta: 0:16:23  lr: 0.000085  grad: 0.1026 (0.1046)  loss: 0.7985 (0.8094)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [41]  [2900/6250]  eta: 0:15:53  lr: 0.000085  grad: 0.1119 (0.1047)  loss: 0.8022 (0.8092)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [41]  [3000/6250]  eta: 0:15:23  lr: 0.000085  grad: 0.1115 (0.1048)  loss: 0.7984 (0.8090)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [41]  [3100/6250]  eta: 0:14:58  lr: 0.000085  grad: 0.1101 (0.1050)  loss: 0.7999 (0.8088)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [41]  [3200/6250]  eta: 0:14:28  lr: 0.000085  grad: 0.1030 (0.1050)  loss: 0.7994 (0.8086)  time: 0.2788  data: 0.0002  max mem: 26157
Train: [41]  [3300/6250]  eta: 0:13:59  lr: 0.000085  grad: 0.1015 (0.1051)  loss: 0.8048 (0.8084)  time: 0.2765  data: 0.0002  max mem: 26157
Train: [41]  [3400/6250]  eta: 0:13:30  lr: 0.000085  grad: 0.1059 (0.1051)  loss: 0.8058 (0.8082)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [41]  [3500/6250]  eta: 0:13:08  lr: 0.000085  grad: 0.1047 (0.1052)  loss: 0.8084 (0.8080)  time: 0.7086  data: 0.4377  max mem: 26157
Train: [41]  [3600/6250]  eta: 0:12:38  lr: 0.000085  grad: 0.1104 (0.1054)  loss: 0.7958 (0.8079)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [41]  [3700/6250]  eta: 0:12:08  lr: 0.000085  grad: 0.1057 (0.1055)  loss: 0.7946 (0.8077)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [41]  [3800/6250]  eta: 0:11:38  lr: 0.000085  grad: 0.1056 (0.1056)  loss: 0.8015 (0.8074)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [41]  [3900/6250]  eta: 0:11:09  lr: 0.000084  grad: 0.1077 (0.1058)  loss: 0.7987 (0.8072)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [41]  [4000/6250]  eta: 0:10:39  lr: 0.000084  grad: 0.1027 (0.1060)  loss: 0.7989 (0.8070)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [41]  [4100/6250]  eta: 0:10:10  lr: 0.000084  grad: 0.1061 (0.1063)  loss: 0.7993 (0.8068)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [41]  [4200/6250]  eta: 0:09:41  lr: 0.000084  grad: 0.1055 (0.1063)  loss: 0.7968 (0.8067)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [41]  [4300/6250]  eta: 0:09:12  lr: 0.000084  grad: 0.0981 (0.1064)  loss: 0.8036 (0.8065)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [41]  [4400/6250]  eta: 0:08:43  lr: 0.000084  grad: 0.1132 (0.1065)  loss: 0.7986 (0.8064)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [41]  [4500/6250]  eta: 0:08:14  lr: 0.000084  grad: 0.1065 (0.1066)  loss: 0.7982 (0.8062)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [41]  [4600/6250]  eta: 0:07:46  lr: 0.000084  grad: 0.1050 (0.1067)  loss: 0.8043 (0.8062)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [41]  [4700/6250]  eta: 0:07:17  lr: 0.000084  grad: 0.1060 (0.1070)  loss: 0.8055 (0.8061)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [41]  [4800/6250]  eta: 0:06:48  lr: 0.000084  grad: 0.1094 (0.1071)  loss: 0.8053 (0.8061)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [41]  [4900/6250]  eta: 0:06:20  lr: 0.000084  grad: 0.1051 (0.1072)  loss: 0.7986 (0.8061)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [41]  [5000/6250]  eta: 0:05:53  lr: 0.000084  grad: 0.1022 (0.1073)  loss: 0.8109 (0.8061)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [41]  [5100/6250]  eta: 0:05:25  lr: 0.000084  grad: 0.1106 (0.1073)  loss: 0.8093 (0.8062)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [41]  [5200/6250]  eta: 0:04:56  lr: 0.000084  grad: 0.0994 (0.1074)  loss: 0.8107 (0.8061)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [41]  [5300/6250]  eta: 0:04:28  lr: 0.000084  grad: 0.1075 (0.1074)  loss: 0.8093 (0.8062)  time: 0.3278  data: 0.0529  max mem: 26157
Train: [41]  [5400/6250]  eta: 0:04:00  lr: 0.000084  grad: 0.1006 (0.1074)  loss: 0.8026 (0.8062)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [41]  [5500/6250]  eta: 0:03:31  lr: 0.000084  grad: 0.1049 (0.1074)  loss: 0.8033 (0.8062)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [41]  [5600/6250]  eta: 0:03:03  lr: 0.000084  grad: 0.1103 (0.1074)  loss: 0.8117 (0.8062)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [41]  [5700/6250]  eta: 0:02:34  lr: 0.000084  grad: 0.1037 (0.1074)  loss: 0.8069 (0.8062)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [41]  [5800/6250]  eta: 0:02:06  lr: 0.000084  grad: 0.1028 (0.1074)  loss: 0.8131 (0.8062)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [41]  [5900/6250]  eta: 0:01:38  lr: 0.000084  grad: 0.1066 (0.1073)  loss: 0.8042 (0.8063)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [41]  [6000/6250]  eta: 0:01:10  lr: 0.000084  grad: 0.1075 (0.1074)  loss: 0.7989 (0.8063)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [41]  [6100/6250]  eta: 0:00:42  lr: 0.000084  grad: 0.1070 (0.1074)  loss: 0.8037 (0.8062)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [41]  [6200/6250]  eta: 0:00:14  lr: 0.000084  grad: 0.1027 (0.1075)  loss: 0.8127 (0.8062)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [41]  [6249/6250]  eta: 0:00:00  lr: 0.000084  grad: 0.1070 (0.1075)  loss: 0.8025 (0.8062)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [41] Total time: 0:29:23 (0.2821 s / it)
Averaged stats: lr: 0.000084  grad: 0.1070 (0.1075)  loss: 0.8025 (0.8062)
Eval (hcp-train-subset): [41]  [ 0/62]  eta: 0:03:25  loss: 0.8341 (0.8341)  time: 3.3181  data: 3.1990  max mem: 26157
Eval (hcp-train-subset): [41]  [61/62]  eta: 0:00:00  loss: 0.8211 (0.8255)  time: 0.1311  data: 0.0461  max mem: 26157
Eval (hcp-train-subset): [41] Total time: 0:00:13 (0.2119 s / it)
Averaged stats (hcp-train-subset): loss: 0.8211 (0.8255)
Making plots (hcp-train-subset): example=52
Eval (hcp-val): [41]  [ 0/62]  eta: 0:05:14  loss: 0.8235 (0.8235)  time: 5.0685  data: 4.9850  max mem: 26157
Eval (hcp-val): [41]  [61/62]  eta: 0:00:00  loss: 0.8253 (0.8266)  time: 0.1280  data: 0.0453  max mem: 26157
Eval (hcp-val): [41] Total time: 0:00:13 (0.2123 s / it)
Averaged stats (hcp-val): loss: 0.8253 (0.8266)
Making plots (hcp-val): example=33
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [42]  [   0/6250]  eta: 6:53:38  lr: 0.000084  grad: 0.0863 (0.0863)  loss: 0.8312 (0.8312)  time: 3.9709  data: 3.6112  max mem: 26157
Train: [42]  [ 100/6250]  eta: 0:33:16  lr: 0.000084  grad: 0.0946 (0.1138)  loss: 0.8339 (0.8336)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [42]  [ 200/6250]  eta: 0:30:02  lr: 0.000084  grad: 0.1013 (0.1138)  loss: 0.8065 (0.8242)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [42]  [ 300/6250]  eta: 0:28:37  lr: 0.000084  grad: 0.0982 (0.1122)  loss: 0.8089 (0.8187)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [42]  [ 400/6250]  eta: 0:27:39  lr: 0.000084  grad: 0.1086 (0.1109)  loss: 0.8097 (0.8152)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [42]  [ 500/6250]  eta: 0:26:57  lr: 0.000084  grad: 0.1050 (0.1102)  loss: 0.8013 (0.8128)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [42]  [ 600/6250]  eta: 0:26:17  lr: 0.000084  grad: 0.1100 (0.1095)  loss: 0.8106 (0.8112)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [42]  [ 700/6250]  eta: 0:26:00  lr: 0.000084  grad: 0.1076 (0.1096)  loss: 0.8058 (0.8099)  time: 0.3675  data: 0.0971  max mem: 26157
Train: [42]  [ 800/6250]  eta: 0:25:40  lr: 0.000084  grad: 0.1118 (0.1097)  loss: 0.7971 (0.8087)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [42]  [ 900/6250]  eta: 0:25:10  lr: 0.000083  grad: 0.1095 (0.1097)  loss: 0.8074 (0.8080)  time: 0.3157  data: 0.0436  max mem: 26157
Train: [42]  [1000/6250]  eta: 0:24:35  lr: 0.000083  grad: 0.1020 (0.1096)  loss: 0.8026 (0.8074)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [42]  [1100/6250]  eta: 0:24:02  lr: 0.000083  grad: 0.1029 (0.1093)  loss: 0.8014 (0.8070)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [42]  [1200/6250]  eta: 0:23:29  lr: 0.000083  grad: 0.1001 (0.1092)  loss: 0.8031 (0.8065)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [42]  [1300/6250]  eta: 0:22:57  lr: 0.000083  grad: 0.1048 (0.1092)  loss: 0.8032 (0.8061)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [42]  [1400/6250]  eta: 0:22:26  lr: 0.000083  grad: 0.1095 (0.1094)  loss: 0.8019 (0.8057)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [42]  [1500/6250]  eta: 0:21:56  lr: 0.000083  grad: 0.1085 (0.1097)  loss: 0.7957 (0.8053)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [42]  [1600/6250]  eta: 0:21:27  lr: 0.000083  grad: 0.1128 (0.1100)  loss: 0.7998 (0.8049)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [42]  [1700/6250]  eta: 0:20:57  lr: 0.000083  grad: 0.1112 (0.1102)  loss: 0.7900 (0.8044)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [42]  [1800/6250]  eta: 0:20:28  lr: 0.000083  grad: 0.1058 (0.1103)  loss: 0.7931 (0.8040)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [42]  [1900/6250]  eta: 0:19:59  lr: 0.000083  grad: 0.1113 (0.1107)  loss: 0.7934 (0.8035)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [42]  [2000/6250]  eta: 0:19:30  lr: 0.000083  grad: 0.1080 (0.1107)  loss: 0.7950 (0.8031)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [42]  [2100/6250]  eta: 0:19:01  lr: 0.000083  grad: 0.1145 (0.1107)  loss: 0.7885 (0.8029)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [42]  [2200/6250]  eta: 0:18:33  lr: 0.000083  grad: 0.1045 (0.1109)  loss: 0.7984 (0.8026)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [42]  [2300/6250]  eta: 0:18:04  lr: 0.000083  grad: 0.1105 (0.1111)  loss: 0.7969 (0.8024)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [42]  [2400/6250]  eta: 0:17:36  lr: 0.000083  grad: 0.1119 (0.1112)  loss: 0.7907 (0.8021)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [42]  [2500/6250]  eta: 0:17:08  lr: 0.000083  grad: 0.1044 (0.1112)  loss: 0.7967 (0.8020)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [42]  [2600/6250]  eta: 0:16:40  lr: 0.000083  grad: 0.1103 (0.1115)  loss: 0.7954 (0.8017)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [42]  [2700/6250]  eta: 0:16:12  lr: 0.000083  grad: 0.1155 (0.1117)  loss: 0.8015 (0.8015)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [42]  [2800/6250]  eta: 0:15:44  lr: 0.000083  grad: 0.1115 (0.1118)  loss: 0.7945 (0.8013)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [42]  [2900/6250]  eta: 0:15:16  lr: 0.000083  grad: 0.1136 (0.1118)  loss: 0.8004 (0.8012)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [42]  [3000/6250]  eta: 0:14:48  lr: 0.000083  grad: 0.1080 (0.1119)  loss: 0.8007 (0.8011)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [42]  [3100/6250]  eta: 0:14:20  lr: 0.000083  grad: 0.1063 (0.1118)  loss: 0.7955 (0.8011)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [42]  [3200/6250]  eta: 0:13:52  lr: 0.000083  grad: 0.1112 (0.1119)  loss: 0.7998 (0.8010)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [42]  [3300/6250]  eta: 0:13:25  lr: 0.000083  grad: 0.1022 (0.1117)  loss: 0.8028 (0.8011)  time: 0.2906  data: 0.0004  max mem: 26157
Train: [42]  [3400/6250]  eta: 0:12:58  lr: 0.000083  grad: 0.1035 (0.1116)  loss: 0.8016 (0.8011)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [42]  [3500/6250]  eta: 0:12:31  lr: 0.000083  grad: 0.1041 (0.1116)  loss: 0.8000 (0.8010)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [42]  [3600/6250]  eta: 0:12:04  lr: 0.000083  grad: 0.1048 (0.1115)  loss: 0.8034 (0.8009)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [42]  [3700/6250]  eta: 0:11:36  lr: 0.000083  grad: 0.0998 (0.1115)  loss: 0.8050 (0.8009)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [42]  [3800/6250]  eta: 0:11:09  lr: 0.000083  grad: 0.1034 (0.1114)  loss: 0.8097 (0.8009)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [42]  [3900/6250]  eta: 0:10:41  lr: 0.000083  grad: 0.1064 (0.1114)  loss: 0.8012 (0.8009)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [42]  [4000/6250]  eta: 0:10:14  lr: 0.000083  grad: 0.1036 (0.1115)  loss: 0.8011 (0.8009)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [42]  [4100/6250]  eta: 0:09:48  lr: 0.000082  grad: 0.1091 (0.1115)  loss: 0.8003 (0.8010)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [42]  [4200/6250]  eta: 0:09:21  lr: 0.000082  grad: 0.1044 (0.1116)  loss: 0.7996 (0.8010)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [42]  [4300/6250]  eta: 0:08:53  lr: 0.000082  grad: 0.1074 (0.1117)  loss: 0.8052 (0.8012)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [42]  [4400/6250]  eta: 0:08:26  lr: 0.000082  grad: 0.1278 (0.1119)  loss: 0.8048 (0.8012)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [42]  [4500/6250]  eta: 0:07:58  lr: 0.000082  grad: 0.1208 (0.1120)  loss: 0.8085 (0.8013)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [42]  [4600/6250]  eta: 0:07:31  lr: 0.000082  grad: 0.1086 (0.1124)  loss: 0.8041 (0.8014)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [42]  [4700/6250]  eta: 0:07:03  lr: 0.000082  grad: 0.1126 (0.1125)  loss: 0.8049 (0.8016)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [42]  [4800/6250]  eta: 0:06:36  lr: 0.000082  grad: 0.1102 (0.1127)  loss: 0.8026 (0.8016)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [42]  [4900/6250]  eta: 0:06:08  lr: 0.000082  grad: 0.1083 (0.1128)  loss: 0.8079 (0.8017)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [42]  [5000/6250]  eta: 0:05:41  lr: 0.000082  grad: 0.1112 (0.1128)  loss: 0.8006 (0.8017)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [42]  [5100/6250]  eta: 0:05:14  lr: 0.000082  grad: 0.1077 (0.1128)  loss: 0.8092 (0.8017)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [42]  [5200/6250]  eta: 0:04:46  lr: 0.000082  grad: 0.1045 (0.1127)  loss: 0.8121 (0.8019)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [42]  [5300/6250]  eta: 0:04:19  lr: 0.000082  grad: 0.1131 (0.1128)  loss: 0.8036 (0.8019)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [42]  [5400/6250]  eta: 0:03:51  lr: 0.000082  grad: 0.1147 (0.1129)  loss: 0.8054 (0.8019)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [42]  [5500/6250]  eta: 0:03:24  lr: 0.000082  grad: 0.1170 (0.1130)  loss: 0.8044 (0.8019)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [42]  [5600/6250]  eta: 0:02:57  lr: 0.000082  grad: 0.1031 (0.1130)  loss: 0.8037 (0.8020)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [42]  [5700/6250]  eta: 0:02:29  lr: 0.000082  grad: 0.1112 (0.1130)  loss: 0.8051 (0.8020)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [42]  [5800/6250]  eta: 0:02:02  lr: 0.000082  grad: 0.1142 (0.1130)  loss: 0.8020 (0.8021)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [42]  [5900/6250]  eta: 0:01:35  lr: 0.000082  grad: 0.1085 (0.1130)  loss: 0.8038 (0.8021)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [42]  [6000/6250]  eta: 0:01:08  lr: 0.000082  grad: 0.1017 (0.1130)  loss: 0.8035 (0.8021)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [42]  [6100/6250]  eta: 0:00:40  lr: 0.000082  grad: 0.1062 (0.1132)  loss: 0.8106 (0.8021)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [42]  [6200/6250]  eta: 0:00:13  lr: 0.000082  grad: 0.1162 (0.1132)  loss: 0.8009 (0.8022)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [42]  [6249/6250]  eta: 0:00:00  lr: 0.000082  grad: 0.1215 (0.1132)  loss: 0.8002 (0.8022)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [42] Total time: 0:28:29 (0.2735 s / it)
Averaged stats: lr: 0.000082  grad: 0.1215 (0.1132)  loss: 0.8002 (0.8022)
Eval (hcp-train-subset): [42]  [ 0/62]  eta: 0:05:03  loss: 0.8303 (0.8303)  time: 4.9029  data: 4.8185  max mem: 26157
Eval (hcp-train-subset): [42]  [61/62]  eta: 0:00:00  loss: 0.8231 (0.8248)  time: 0.1277  data: 0.0426  max mem: 26157
Eval (hcp-train-subset): [42] Total time: 0:00:13 (0.2235 s / it)
Averaged stats (hcp-train-subset): loss: 0.8231 (0.8248)
Making plots (hcp-train-subset): example=46
Eval (hcp-val): [42]  [ 0/62]  eta: 0:05:54  loss: 0.8228 (0.8228)  time: 5.7166  data: 5.6322  max mem: 26157
Eval (hcp-val): [42]  [61/62]  eta: 0:00:00  loss: 0.8263 (0.8270)  time: 0.1393  data: 0.0543  max mem: 26157
Eval (hcp-val): [42] Total time: 0:00:13 (0.2187 s / it)
Averaged stats (hcp-val): loss: 0.8263 (0.8270)
Making plots (hcp-val): example=45
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [43]  [   0/6250]  eta: 9:07:24  lr: 0.000082  grad: 0.0717 (0.0717)  loss: 0.8529 (0.8529)  time: 5.2550  data: 4.8607  max mem: 26157
Train: [43]  [ 100/6250]  eta: 0:33:43  lr: 0.000082  grad: 0.1010 (0.1140)  loss: 0.8145 (0.8287)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [43]  [ 200/6250]  eta: 0:30:13  lr: 0.000082  grad: 0.1044 (0.1145)  loss: 0.8003 (0.8200)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [43]  [ 300/6250]  eta: 0:28:46  lr: 0.000082  grad: 0.1094 (0.1141)  loss: 0.8076 (0.8162)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [43]  [ 400/6250]  eta: 0:27:48  lr: 0.000082  grad: 0.1053 (0.1130)  loss: 0.8034 (0.8139)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [43]  [ 500/6250]  eta: 0:27:03  lr: 0.000082  grad: 0.1059 (0.1119)  loss: 0.8057 (0.8127)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [43]  [ 600/6250]  eta: 0:26:25  lr: 0.000082  grad: 0.0984 (0.1107)  loss: 0.8100 (0.8120)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [43]  [ 700/6250]  eta: 0:25:49  lr: 0.000082  grad: 0.1052 (0.1098)  loss: 0.8094 (0.8113)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [43]  [ 800/6250]  eta: 0:25:15  lr: 0.000082  grad: 0.1089 (0.1094)  loss: 0.8168 (0.8112)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [43]  [ 900/6250]  eta: 0:24:44  lr: 0.000082  grad: 0.1021 (0.1096)  loss: 0.8034 (0.8109)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [43]  [1000/6250]  eta: 0:24:14  lr: 0.000081  grad: 0.1066 (0.1091)  loss: 0.8081 (0.8107)  time: 0.2764  data: 0.0002  max mem: 26157
Train: [43]  [1100/6250]  eta: 0:23:48  lr: 0.000081  grad: 0.1022 (0.1088)  loss: 0.8106 (0.8107)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [43]  [1200/6250]  eta: 0:23:16  lr: 0.000081  grad: 0.1132 (0.1087)  loss: 0.8051 (0.8103)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [43]  [1300/6250]  eta: 0:22:46  lr: 0.000081  grad: 0.0952 (0.1087)  loss: 0.8075 (0.8102)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [43]  [1400/6250]  eta: 0:22:18  lr: 0.000081  grad: 0.1016 (0.1086)  loss: 0.8097 (0.8099)  time: 0.2734  data: 0.0003  max mem: 26157
Train: [43]  [1500/6250]  eta: 0:22:04  lr: 0.000081  grad: 0.1046 (0.1084)  loss: 0.8079 (0.8097)  time: 0.3942  data: 0.1238  max mem: 26157
Train: [43]  [1600/6250]  eta: 0:21:34  lr: 0.000081  grad: 0.0980 (0.1083)  loss: 0.8062 (0.8094)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [43]  [1700/6250]  eta: 0:21:04  lr: 0.000081  grad: 0.1037 (0.1083)  loss: 0.8057 (0.8093)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [43]  [1800/6250]  eta: 0:20:34  lr: 0.000081  grad: 0.1112 (0.1084)  loss: 0.8035 (0.8092)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [43]  [1900/6250]  eta: 0:20:04  lr: 0.000081  grad: 0.1032 (0.1083)  loss: 0.8045 (0.8090)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [43]  [2000/6250]  eta: 0:19:35  lr: 0.000081  grad: 0.1029 (0.1081)  loss: 0.8012 (0.8090)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [43]  [2100/6250]  eta: 0:19:06  lr: 0.000081  grad: 0.1032 (0.1081)  loss: 0.8108 (0.8089)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [43]  [2200/6250]  eta: 0:18:37  lr: 0.000081  grad: 0.0976 (0.1081)  loss: 0.8127 (0.8089)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [43]  [2300/6250]  eta: 0:18:08  lr: 0.000081  grad: 0.1058 (0.1079)  loss: 0.8094 (0.8089)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [43]  [2400/6250]  eta: 0:17:40  lr: 0.000081  grad: 0.1113 (0.1083)  loss: 0.8078 (0.8088)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [43]  [2500/6250]  eta: 0:17:25  lr: 0.000081  grad: 0.1101 (0.1083)  loss: 0.8093 (0.8089)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [43]  [2600/6250]  eta: 0:16:56  lr: 0.000081  grad: 0.1089 (0.1083)  loss: 0.8012 (0.8088)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [43]  [2700/6250]  eta: 0:16:27  lr: 0.000081  grad: 0.1150 (0.1083)  loss: 0.8096 (0.8087)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [43]  [2800/6250]  eta: 0:15:58  lr: 0.000081  grad: 0.1033 (0.1085)  loss: 0.8049 (0.8086)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [43]  [2900/6250]  eta: 0:15:29  lr: 0.000081  grad: 0.1156 (0.1087)  loss: 0.8092 (0.8085)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [43]  [3000/6250]  eta: 0:15:00  lr: 0.000081  grad: 0.1073 (0.1091)  loss: 0.8022 (0.8085)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [43]  [3100/6250]  eta: 0:14:32  lr: 0.000081  grad: 0.1072 (0.1092)  loss: 0.8086 (0.8084)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [43]  [3200/6250]  eta: 0:14:03  lr: 0.000081  grad: 0.1092 (0.1093)  loss: 0.7957 (0.8083)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [43]  [3300/6250]  eta: 0:13:35  lr: 0.000081  grad: 0.1136 (0.1096)  loss: 0.7984 (0.8082)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [43]  [3400/6250]  eta: 0:13:07  lr: 0.000081  grad: 0.1121 (0.1098)  loss: 0.8114 (0.8081)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [43]  [3500/6250]  eta: 0:12:38  lr: 0.000081  grad: 0.1066 (0.1098)  loss: 0.8052 (0.8079)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [43]  [3600/6250]  eta: 0:12:10  lr: 0.000081  grad: 0.1133 (0.1099)  loss: 0.8026 (0.8077)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [43]  [3700/6250]  eta: 0:11:42  lr: 0.000081  grad: 0.1035 (0.1100)  loss: 0.8052 (0.8076)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [43]  [3800/6250]  eta: 0:11:14  lr: 0.000081  grad: 0.1108 (0.1101)  loss: 0.8048 (0.8075)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [43]  [3900/6250]  eta: 0:10:46  lr: 0.000081  grad: 0.1119 (0.1101)  loss: 0.7954 (0.8074)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [43]  [4000/6250]  eta: 0:10:19  lr: 0.000081  grad: 0.1083 (0.1102)  loss: 0.8050 (0.8072)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [43]  [4100/6250]  eta: 0:09:51  lr: 0.000081  grad: 0.1113 (0.1104)  loss: 0.7936 (0.8071)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [43]  [4200/6250]  eta: 0:09:23  lr: 0.000080  grad: 0.1098 (0.1105)  loss: 0.8084 (0.8071)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [43]  [4300/6250]  eta: 0:08:55  lr: 0.000080  grad: 0.1113 (0.1105)  loss: 0.7944 (0.8070)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [43]  [4400/6250]  eta: 0:08:28  lr: 0.000080  grad: 0.1127 (0.1105)  loss: 0.8083 (0.8069)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [43]  [4500/6250]  eta: 0:08:00  lr: 0.000080  grad: 0.1079 (0.1106)  loss: 0.7990 (0.8068)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [43]  [4600/6250]  eta: 0:07:32  lr: 0.000080  grad: 0.1097 (0.1106)  loss: 0.8020 (0.8066)  time: 0.2767  data: 0.0002  max mem: 26157
Train: [43]  [4700/6250]  eta: 0:07:05  lr: 0.000080  grad: 0.1142 (0.1108)  loss: 0.8017 (0.8066)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [43]  [4800/6250]  eta: 0:06:39  lr: 0.000080  grad: 0.1030 (0.1108)  loss: 0.8044 (0.8065)  time: 0.2741  data: 0.0002  max mem: 26157
Train: [43]  [4900/6250]  eta: 0:06:11  lr: 0.000080  grad: 0.1043 (0.1108)  loss: 0.8072 (0.8065)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [43]  [5000/6250]  eta: 0:05:44  lr: 0.000080  grad: 0.1091 (0.1109)  loss: 0.8012 (0.8065)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [43]  [5100/6250]  eta: 0:05:16  lr: 0.000080  grad: 0.1006 (0.1109)  loss: 0.8076 (0.8065)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [43]  [5200/6250]  eta: 0:04:49  lr: 0.000080  grad: 0.1081 (0.1109)  loss: 0.8013 (0.8065)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [43]  [5300/6250]  eta: 0:04:21  lr: 0.000080  grad: 0.1079 (0.1109)  loss: 0.8017 (0.8065)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [43]  [5400/6250]  eta: 0:03:53  lr: 0.000080  grad: 0.1116 (0.1109)  loss: 0.8088 (0.8065)  time: 0.2736  data: 0.0002  max mem: 26157
Train: [43]  [5500/6250]  eta: 0:03:26  lr: 0.000080  grad: 0.1066 (0.1110)  loss: 0.8059 (0.8065)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [43]  [5600/6250]  eta: 0:02:58  lr: 0.000080  grad: 0.1034 (0.1111)  loss: 0.8097 (0.8065)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [43]  [5700/6250]  eta: 0:02:31  lr: 0.000080  grad: 0.1000 (0.1111)  loss: 0.8158 (0.8065)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [43]  [5800/6250]  eta: 0:02:03  lr: 0.000080  grad: 0.1083 (0.1110)  loss: 0.8021 (0.8065)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [43]  [5900/6250]  eta: 0:01:36  lr: 0.000080  grad: 0.1120 (0.1110)  loss: 0.8049 (0.8065)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [43]  [6000/6250]  eta: 0:01:09  lr: 0.000080  grad: 0.1060 (0.1109)  loss: 0.8097 (0.8065)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [43]  [6100/6250]  eta: 0:00:41  lr: 0.000080  grad: 0.1049 (0.1109)  loss: 0.7984 (0.8065)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [43]  [6200/6250]  eta: 0:00:13  lr: 0.000080  grad: 0.1037 (0.1109)  loss: 0.8089 (0.8065)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [43]  [6249/6250]  eta: 0:00:00  lr: 0.000080  grad: 0.1063 (0.1109)  loss: 0.8090 (0.8065)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [43] Total time: 0:28:56 (0.2778 s / it)
Averaged stats: lr: 0.000080  grad: 0.1063 (0.1109)  loss: 0.8090 (0.8065)
Eval (hcp-train-subset): [43]  [ 0/62]  eta: 0:03:55  loss: 0.8356 (0.8356)  time: 3.8025  data: 3.6804  max mem: 26157
Eval (hcp-train-subset): [43]  [61/62]  eta: 0:00:00  loss: 0.8210 (0.8229)  time: 0.1047  data: 0.0221  max mem: 26157
Eval (hcp-train-subset): [43] Total time: 0:00:12 (0.2007 s / it)
Averaged stats (hcp-train-subset): loss: 0.8210 (0.8229)
Making plots (hcp-train-subset): example=29
Eval (hcp-val): [43]  [ 0/62]  eta: 0:04:54  loss: 0.8209 (0.8209)  time: 4.7487  data: 4.6639  max mem: 26157
Eval (hcp-val): [43]  [61/62]  eta: 0:00:00  loss: 0.8238 (0.8264)  time: 0.1347  data: 0.0503  max mem: 26157
Eval (hcp-val): [43] Total time: 0:00:13 (0.2100 s / it)
Averaged stats (hcp-val): loss: 0.8238 (0.8264)
Making plots (hcp-val): example=51
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [44]  [   0/6250]  eta: 10:06:22  lr: 0.000080  grad: 0.2292 (0.2292)  loss: 0.8479 (0.8479)  time: 5.8212  data: 5.5185  max mem: 26157
Train: [44]  [ 100/6250]  eta: 0:34:45  lr: 0.000080  grad: 0.0970 (0.1264)  loss: 0.8311 (0.8178)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [44]  [ 200/6250]  eta: 0:30:43  lr: 0.000080  grad: 0.1115 (0.1288)  loss: 0.8016 (0.8096)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [44]  [ 300/6250]  eta: 0:29:07  lr: 0.000080  grad: 0.1143 (0.1259)  loss: 0.7938 (0.8057)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [44]  [ 400/6250]  eta: 0:28:03  lr: 0.000080  grad: 0.1179 (0.1257)  loss: 0.8002 (0.8042)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [44]  [ 500/6250]  eta: 0:27:13  lr: 0.000080  grad: 0.1070 (0.1237)  loss: 0.8034 (0.8036)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [44]  [ 600/6250]  eta: 0:26:32  lr: 0.000080  grad: 0.1041 (0.1225)  loss: 0.8029 (0.8036)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [44]  [ 700/6250]  eta: 0:25:53  lr: 0.000080  grad: 0.1134 (0.1214)  loss: 0.8062 (0.8033)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [44]  [ 800/6250]  eta: 0:25:19  lr: 0.000080  grad: 0.1039 (0.1201)  loss: 0.8002 (0.8027)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [44]  [ 900/6250]  eta: 0:24:46  lr: 0.000080  grad: 0.1150 (0.1194)  loss: 0.7990 (0.8025)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [44]  [1000/6250]  eta: 0:24:14  lr: 0.000080  grad: 0.1080 (0.1185)  loss: 0.7996 (0.8027)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [44]  [1100/6250]  eta: 0:23:43  lr: 0.000079  grad: 0.1086 (0.1177)  loss: 0.8055 (0.8027)  time: 0.2702  data: 0.0003  max mem: 26157
Train: [44]  [1200/6250]  eta: 0:23:13  lr: 0.000079  grad: 0.1038 (0.1171)  loss: 0.8092 (0.8027)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [44]  [1300/6250]  eta: 0:23:01  lr: 0.000079  grad: 0.1024 (0.1166)  loss: 0.7960 (0.8027)  time: 0.2751  data: 0.0003  max mem: 26157
Train: [44]  [1400/6250]  eta: 0:22:32  lr: 0.000079  grad: 0.1058 (0.1164)  loss: 0.8113 (0.8025)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [44]  [1500/6250]  eta: 0:22:01  lr: 0.000079  grad: 0.1123 (0.1163)  loss: 0.7994 (0.8024)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [44]  [1600/6250]  eta: 0:21:31  lr: 0.000079  grad: 0.1106 (0.1163)  loss: 0.8003 (0.8022)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [44]  [1700/6250]  eta: 0:21:02  lr: 0.000079  grad: 0.1168 (0.1164)  loss: 0.7949 (0.8020)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [44]  [1800/6250]  eta: 0:20:45  lr: 0.000079  grad: 0.1160 (0.1164)  loss: 0.7986 (0.8018)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [44]  [1900/6250]  eta: 0:20:15  lr: 0.000079  grad: 0.1105 (0.1164)  loss: 0.7983 (0.8017)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [44]  [2000/6250]  eta: 0:19:45  lr: 0.000079  grad: 0.1100 (0.1162)  loss: 0.8080 (0.8016)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [44]  [2100/6250]  eta: 0:19:15  lr: 0.000079  grad: 0.1213 (0.1162)  loss: 0.8011 (0.8016)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [44]  [2200/6250]  eta: 0:18:46  lr: 0.000079  grad: 0.1065 (0.1161)  loss: 0.8043 (0.8016)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [44]  [2300/6250]  eta: 0:18:16  lr: 0.000079  grad: 0.1022 (0.1158)  loss: 0.8029 (0.8017)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [44]  [2400/6250]  eta: 0:17:47  lr: 0.000079  grad: 0.1077 (0.1157)  loss: 0.8039 (0.8018)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [44]  [2500/6250]  eta: 0:17:18  lr: 0.000079  grad: 0.1067 (0.1155)  loss: 0.8028 (0.8019)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [44]  [2600/6250]  eta: 0:16:52  lr: 0.000079  grad: 0.1032 (0.1152)  loss: 0.7983 (0.8019)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [44]  [2700/6250]  eta: 0:16:29  lr: 0.000079  grad: 0.1091 (0.1150)  loss: 0.7987 (0.8019)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [44]  [2800/6250]  eta: 0:16:01  lr: 0.000079  grad: 0.1169 (0.1153)  loss: 0.8024 (0.8020)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [44]  [2900/6250]  eta: 0:15:32  lr: 0.000079  grad: 0.1101 (0.1151)  loss: 0.8027 (0.8021)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [44]  [3000/6250]  eta: 0:15:03  lr: 0.000079  grad: 0.1062 (0.1151)  loss: 0.8051 (0.8021)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [44]  [3100/6250]  eta: 0:14:34  lr: 0.000079  grad: 0.1102 (0.1151)  loss: 0.8074 (0.8022)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [44]  [3200/6250]  eta: 0:14:06  lr: 0.000079  grad: 0.1068 (0.1151)  loss: 0.8091 (0.8024)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [44]  [3300/6250]  eta: 0:13:37  lr: 0.000079  grad: 0.1088 (0.1150)  loss: 0.8084 (0.8025)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [44]  [3400/6250]  eta: 0:13:09  lr: 0.000079  grad: 0.1010 (0.1150)  loss: 0.8038 (0.8026)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [44]  [3500/6250]  eta: 0:12:41  lr: 0.000079  grad: 0.1148 (0.1149)  loss: 0.7980 (0.8026)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [44]  [3600/6250]  eta: 0:12:12  lr: 0.000079  grad: 0.1068 (0.1149)  loss: 0.8098 (0.8027)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [44]  [3700/6250]  eta: 0:11:44  lr: 0.000079  grad: 0.1149 (0.1149)  loss: 0.8035 (0.8028)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [44]  [3800/6250]  eta: 0:11:16  lr: 0.000079  grad: 0.1061 (0.1148)  loss: 0.8043 (0.8028)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [44]  [3900/6250]  eta: 0:10:48  lr: 0.000079  grad: 0.1076 (0.1150)  loss: 0.7977 (0.8027)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [44]  [4000/6250]  eta: 0:10:20  lr: 0.000079  grad: 0.1072 (0.1149)  loss: 0.8025 (0.8027)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [44]  [4100/6250]  eta: 0:09:52  lr: 0.000079  grad: 0.1114 (0.1149)  loss: 0.7978 (0.8026)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [44]  [4200/6250]  eta: 0:09:24  lr: 0.000078  grad: 0.1053 (0.1148)  loss: 0.8002 (0.8026)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [44]  [4300/6250]  eta: 0:08:57  lr: 0.000078  grad: 0.1124 (0.1148)  loss: 0.7981 (0.8026)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [44]  [4400/6250]  eta: 0:08:29  lr: 0.000078  grad: 0.1179 (0.1148)  loss: 0.7929 (0.8025)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [44]  [4500/6250]  eta: 0:08:01  lr: 0.000078  grad: 0.1100 (0.1149)  loss: 0.8012 (0.8025)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [44]  [4600/6250]  eta: 0:07:33  lr: 0.000078  grad: 0.1114 (0.1148)  loss: 0.8043 (0.8024)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [44]  [4700/6250]  eta: 0:07:06  lr: 0.000078  grad: 0.1025 (0.1148)  loss: 0.8031 (0.8023)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [44]  [4800/6250]  eta: 0:06:38  lr: 0.000078  grad: 0.1137 (0.1148)  loss: 0.7939 (0.8022)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [44]  [4900/6250]  eta: 0:06:10  lr: 0.000078  grad: 0.1016 (0.1148)  loss: 0.8059 (0.8022)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [44]  [5000/6250]  eta: 0:05:43  lr: 0.000078  grad: 0.1132 (0.1147)  loss: 0.8011 (0.8022)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [44]  [5100/6250]  eta: 0:05:15  lr: 0.000078  grad: 0.1151 (0.1147)  loss: 0.8047 (0.8022)  time: 0.2726  data: 0.0003  max mem: 26157
Train: [44]  [5200/6250]  eta: 0:04:48  lr: 0.000078  grad: 0.1084 (0.1147)  loss: 0.8043 (0.8022)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [44]  [5300/6250]  eta: 0:04:20  lr: 0.000078  grad: 0.1160 (0.1147)  loss: 0.8009 (0.8022)  time: 0.2695  data: 0.0001  max mem: 26157
Train: [44]  [5400/6250]  eta: 0:03:53  lr: 0.000078  grad: 0.1116 (0.1148)  loss: 0.7998 (0.8022)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [44]  [5500/6250]  eta: 0:03:25  lr: 0.000078  grad: 0.1115 (0.1150)  loss: 0.8015 (0.8022)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [44]  [5600/6250]  eta: 0:02:58  lr: 0.000078  grad: 0.1098 (0.1152)  loss: 0.8045 (0.8022)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [44]  [5700/6250]  eta: 0:02:30  lr: 0.000078  grad: 0.1118 (0.1152)  loss: 0.8101 (0.8022)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [44]  [5800/6250]  eta: 0:02:03  lr: 0.000078  grad: 0.1088 (0.1153)  loss: 0.8110 (0.8022)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [44]  [5900/6250]  eta: 0:01:35  lr: 0.000078  grad: 0.1121 (0.1154)  loss: 0.8047 (0.8022)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [44]  [6000/6250]  eta: 0:01:08  lr: 0.000078  grad: 0.1148 (0.1155)  loss: 0.7967 (0.8021)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [44]  [6100/6250]  eta: 0:00:41  lr: 0.000078  grad: 0.1246 (0.1156)  loss: 0.7956 (0.8021)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [44]  [6200/6250]  eta: 0:00:13  lr: 0.000078  grad: 0.1159 (0.1157)  loss: 0.7963 (0.8020)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [44]  [6249/6250]  eta: 0:00:00  lr: 0.000078  grad: 0.1121 (0.1157)  loss: 0.8011 (0.8020)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [44] Total time: 0:28:37 (0.2748 s / it)
Averaged stats: lr: 0.000078  grad: 0.1121 (0.1157)  loss: 0.8011 (0.8020)
Eval (hcp-train-subset): [44]  [ 0/62]  eta: 0:05:40  loss: 0.8297 (0.8297)  time: 5.4899  data: 5.4064  max mem: 26157
Eval (hcp-train-subset): [44]  [61/62]  eta: 0:00:00  loss: 0.8246 (0.8264)  time: 0.1250  data: 0.0423  max mem: 26157
Eval (hcp-train-subset): [44] Total time: 0:00:13 (0.2181 s / it)
Averaged stats (hcp-train-subset): loss: 0.8246 (0.8264)
Making plots (hcp-train-subset): example=7
Eval (hcp-val): [44]  [ 0/62]  eta: 0:06:02  loss: 0.8213 (0.8213)  time: 5.8437  data: 5.7581  max mem: 26157
Eval (hcp-val): [44]  [61/62]  eta: 0:00:00  loss: 0.8259 (0.8269)  time: 0.1109  data: 0.0285  max mem: 26157
Eval (hcp-val): [44] Total time: 0:00:13 (0.2151 s / it)
Averaged stats (hcp-val): loss: 0.8259 (0.8269)
Making plots (hcp-val): example=6
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [45]  [   0/6250]  eta: 12:55:58  lr: 0.000078  grad: 0.1865 (0.1865)  loss: 0.7921 (0.7921)  time: 7.4493  data: 7.1706  max mem: 26157
Train: [45]  [ 100/6250]  eta: 0:36:49  lr: 0.000078  grad: 0.1300 (0.1671)  loss: 0.8006 (0.8084)  time: 0.3486  data: 0.0769  max mem: 26157
Train: [45]  [ 200/6250]  eta: 0:31:51  lr: 0.000078  grad: 0.1184 (0.1458)  loss: 0.8007 (0.8027)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [45]  [ 300/6250]  eta: 0:29:50  lr: 0.000078  grad: 0.1129 (0.1354)  loss: 0.7982 (0.8024)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [45]  [ 400/6250]  eta: 0:29:38  lr: 0.000078  grad: 0.1088 (0.1298)  loss: 0.8114 (0.8035)  time: 0.4759  data: 0.2028  max mem: 26157
Train: [45]  [ 500/6250]  eta: 0:28:28  lr: 0.000078  grad: 0.1128 (0.1261)  loss: 0.8066 (0.8043)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [45]  [ 600/6250]  eta: 0:28:34  lr: 0.000078  grad: 0.1094 (0.1242)  loss: 0.8076 (0.8044)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [45]  [ 700/6250]  eta: 0:28:06  lr: 0.000078  grad: 0.1021 (0.1221)  loss: 0.8011 (0.8047)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [45]  [ 800/6250]  eta: 0:27:43  lr: 0.000078  grad: 0.1056 (0.1209)  loss: 0.8068 (0.8047)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [45]  [ 900/6250]  eta: 0:27:15  lr: 0.000078  grad: 0.1029 (0.1194)  loss: 0.8085 (0.8050)  time: 0.2808  data: 0.0003  max mem: 26157
Train: [45]  [1000/6250]  eta: 0:26:42  lr: 0.000078  grad: 0.1062 (0.1184)  loss: 0.8055 (0.8053)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [45]  [1100/6250]  eta: 0:25:55  lr: 0.000077  grad: 0.1008 (0.1174)  loss: 0.8104 (0.8055)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [45]  [1200/6250]  eta: 0:25:11  lr: 0.000077  grad: 0.1175 (0.1170)  loss: 0.8095 (0.8055)  time: 0.2720  data: 0.0004  max mem: 26157
Train: [45]  [1300/6250]  eta: 0:24:30  lr: 0.000077  grad: 0.1119 (0.1171)  loss: 0.8056 (0.8054)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [45]  [1400/6250]  eta: 0:23:51  lr: 0.000077  grad: 0.1087 (0.1172)  loss: 0.7947 (0.8051)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [45]  [1500/6250]  eta: 0:23:13  lr: 0.000077  grad: 0.1162 (0.1171)  loss: 0.8020 (0.8050)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [45]  [1600/6250]  eta: 0:22:37  lr: 0.000077  grad: 0.1118 (0.1173)  loss: 0.7984 (0.8049)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [45]  [1700/6250]  eta: 0:22:02  lr: 0.000077  grad: 0.1064 (0.1171)  loss: 0.8017 (0.8048)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [45]  [1800/6250]  eta: 0:21:28  lr: 0.000077  grad: 0.1089 (0.1170)  loss: 0.8027 (0.8048)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [45]  [1900/6250]  eta: 0:20:55  lr: 0.000077  grad: 0.1094 (0.1168)  loss: 0.8077 (0.8047)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [45]  [2000/6250]  eta: 0:20:22  lr: 0.000077  grad: 0.1163 (0.1169)  loss: 0.7998 (0.8046)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [45]  [2100/6250]  eta: 0:19:50  lr: 0.000077  grad: 0.1066 (0.1168)  loss: 0.7987 (0.8045)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [45]  [2200/6250]  eta: 0:19:18  lr: 0.000077  grad: 0.1131 (0.1168)  loss: 0.8019 (0.8045)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [45]  [2300/6250]  eta: 0:18:46  lr: 0.000077  grad: 0.1093 (0.1170)  loss: 0.8077 (0.8043)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [45]  [2400/6250]  eta: 0:18:15  lr: 0.000077  grad: 0.1074 (0.1170)  loss: 0.8078 (0.8042)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [45]  [2500/6250]  eta: 0:17:44  lr: 0.000077  grad: 0.1038 (0.1173)  loss: 0.8070 (0.8041)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [45]  [2600/6250]  eta: 0:17:14  lr: 0.000077  grad: 0.1064 (0.1172)  loss: 0.8027 (0.8041)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [45]  [2700/6250]  eta: 0:16:49  lr: 0.000077  grad: 0.1092 (0.1173)  loss: 0.8029 (0.8041)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [45]  [2800/6250]  eta: 0:16:19  lr: 0.000077  grad: 0.1084 (0.1172)  loss: 0.8108 (0.8041)  time: 0.2995  data: 0.0216  max mem: 26157
Train: [45]  [2900/6250]  eta: 0:15:49  lr: 0.000077  grad: 0.1186 (0.1172)  loss: 0.7899 (0.8040)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [45]  [3000/6250]  eta: 0:15:20  lr: 0.000077  grad: 0.1167 (0.1174)  loss: 0.7979 (0.8039)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [45]  [3100/6250]  eta: 0:14:55  lr: 0.000077  grad: 0.1080 (0.1172)  loss: 0.8040 (0.8039)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [45]  [3200/6250]  eta: 0:14:25  lr: 0.000077  grad: 0.1061 (0.1171)  loss: 0.7986 (0.8039)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [45]  [3300/6250]  eta: 0:13:55  lr: 0.000077  grad: 0.1140 (0.1174)  loss: 0.8003 (0.8039)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [45]  [3400/6250]  eta: 0:13:26  lr: 0.000077  grad: 0.1078 (0.1174)  loss: 0.8021 (0.8038)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [45]  [3500/6250]  eta: 0:12:56  lr: 0.000077  grad: 0.1061 (0.1172)  loss: 0.8048 (0.8038)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [45]  [3600/6250]  eta: 0:12:27  lr: 0.000077  grad: 0.1110 (0.1173)  loss: 0.8044 (0.8038)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [45]  [3700/6250]  eta: 0:11:58  lr: 0.000077  grad: 0.1156 (0.1173)  loss: 0.8046 (0.8038)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [45]  [3800/6250]  eta: 0:11:30  lr: 0.000077  grad: 0.1156 (0.1172)  loss: 0.7978 (0.8037)  time: 0.2739  data: 0.0002  max mem: 26157
Train: [45]  [3900/6250]  eta: 0:11:01  lr: 0.000077  grad: 0.1151 (0.1172)  loss: 0.7937 (0.8036)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [45]  [4000/6250]  eta: 0:10:32  lr: 0.000077  grad: 0.1127 (0.1173)  loss: 0.7965 (0.8035)  time: 0.2715  data: 0.0003  max mem: 26157
Train: [45]  [4100/6250]  eta: 0:10:04  lr: 0.000077  grad: 0.1117 (0.1173)  loss: 0.8059 (0.8035)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [45]  [4200/6250]  eta: 0:09:35  lr: 0.000076  grad: 0.1109 (0.1176)  loss: 0.7994 (0.8034)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [45]  [4300/6250]  eta: 0:09:06  lr: 0.000076  grad: 0.1175 (0.1175)  loss: 0.7980 (0.8033)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [45]  [4400/6250]  eta: 0:08:38  lr: 0.000076  grad: 0.1109 (0.1177)  loss: 0.7992 (0.8032)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [45]  [4500/6250]  eta: 0:08:09  lr: 0.000076  grad: 0.1078 (0.1178)  loss: 0.8037 (0.8032)  time: 0.2738  data: 0.0002  max mem: 26157
Train: [45]  [4600/6250]  eta: 0:07:43  lr: 0.000076  grad: 0.1198 (0.1180)  loss: 0.7979 (0.8032)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [45]  [4700/6250]  eta: 0:07:14  lr: 0.000076  grad: 0.1118 (0.1181)  loss: 0.8069 (0.8031)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [45]  [4800/6250]  eta: 0:06:46  lr: 0.000076  grad: 0.1241 (0.1183)  loss: 0.8013 (0.8030)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [45]  [4900/6250]  eta: 0:06:18  lr: 0.000076  grad: 0.1250 (0.1185)  loss: 0.8048 (0.8029)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [45]  [5000/6250]  eta: 0:05:50  lr: 0.000076  grad: 0.1136 (0.1185)  loss: 0.8018 (0.8029)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [45]  [5100/6250]  eta: 0:05:22  lr: 0.000076  grad: 0.1155 (0.1184)  loss: 0.8025 (0.8028)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [45]  [5200/6250]  eta: 0:04:54  lr: 0.000076  grad: 0.1163 (0.1186)  loss: 0.7996 (0.8028)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [45]  [5300/6250]  eta: 0:04:26  lr: 0.000076  grad: 0.1159 (0.1187)  loss: 0.8039 (0.8027)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [45]  [5400/6250]  eta: 0:03:58  lr: 0.000076  grad: 0.1072 (0.1188)  loss: 0.8017 (0.8026)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [45]  [5500/6250]  eta: 0:03:30  lr: 0.000076  grad: 0.1125 (0.1190)  loss: 0.8019 (0.8026)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [45]  [5600/6250]  eta: 0:03:01  lr: 0.000076  grad: 0.1138 (0.1189)  loss: 0.8032 (0.8026)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [45]  [5700/6250]  eta: 0:02:33  lr: 0.000076  grad: 0.1112 (0.1189)  loss: 0.8032 (0.8025)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [45]  [5800/6250]  eta: 0:02:05  lr: 0.000076  grad: 0.1156 (0.1189)  loss: 0.8032 (0.8025)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [45]  [5900/6250]  eta: 0:01:37  lr: 0.000076  grad: 0.1184 (0.1189)  loss: 0.8000 (0.8025)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [45]  [6000/6250]  eta: 0:01:09  lr: 0.000076  grad: 0.1125 (0.1188)  loss: 0.8055 (0.8025)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [45]  [6100/6250]  eta: 0:00:41  lr: 0.000076  grad: 0.1127 (0.1192)  loss: 0.8011 (0.8025)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [45]  [6200/6250]  eta: 0:00:13  lr: 0.000076  grad: 0.1185 (0.1192)  loss: 0.7963 (0.8024)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [45]  [6249/6250]  eta: 0:00:00  lr: 0.000076  grad: 0.1209 (0.1193)  loss: 0.7982 (0.8024)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [45] Total time: 0:29:11 (0.2803 s / it)
Averaged stats: lr: 0.000076  grad: 0.1209 (0.1193)  loss: 0.7982 (0.8024)
Eval (hcp-train-subset): [45]  [ 0/62]  eta: 0:05:09  loss: 0.8372 (0.8372)  time: 4.9903  data: 4.9063  max mem: 26157
Eval (hcp-train-subset): [45]  [61/62]  eta: 0:00:00  loss: 0.8221 (0.8250)  time: 0.1300  data: 0.0473  max mem: 26157
Eval (hcp-train-subset): [45] Total time: 0:00:14 (0.2330 s / it)
Averaged stats (hcp-train-subset): loss: 0.8221 (0.8250)
Making plots (hcp-train-subset): example=8
Eval (hcp-val): [45]  [ 0/62]  eta: 0:04:48  loss: 0.8210 (0.8210)  time: 4.6591  data: 4.5741  max mem: 26157
Eval (hcp-val): [45]  [61/62]  eta: 0:00:00  loss: 0.8243 (0.8258)  time: 0.1366  data: 0.0532  max mem: 26157
Eval (hcp-val): [45] Total time: 0:00:13 (0.2208 s / it)
Averaged stats (hcp-val): loss: 0.8243 (0.8258)
Making plots (hcp-val): example=27
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [46]  [   0/6250]  eta: 11:06:54  lr: 0.000076  grad: 0.3525 (0.3525)  loss: 0.8067 (0.8067)  time: 6.4024  data: 6.1155  max mem: 26157
Train: [46]  [ 100/6250]  eta: 0:34:29  lr: 0.000076  grad: 0.1238 (0.1352)  loss: 0.8213 (0.8261)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [46]  [ 200/6250]  eta: 0:30:32  lr: 0.000076  grad: 0.1084 (0.1298)  loss: 0.8104 (0.8171)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [46]  [ 300/6250]  eta: 0:28:55  lr: 0.000076  grad: 0.1113 (0.1253)  loss: 0.8112 (0.8131)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [46]  [ 400/6250]  eta: 0:27:52  lr: 0.000076  grad: 0.1011 (0.1207)  loss: 0.8111 (0.8122)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [46]  [ 500/6250]  eta: 0:27:06  lr: 0.000076  grad: 0.1088 (0.1188)  loss: 0.8140 (0.8115)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [46]  [ 600/6250]  eta: 0:26:26  lr: 0.000076  grad: 0.1102 (0.1173)  loss: 0.8137 (0.8110)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [46]  [ 700/6250]  eta: 0:25:52  lr: 0.000076  grad: 0.1055 (0.1167)  loss: 0.8110 (0.8106)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [46]  [ 800/6250]  eta: 0:25:19  lr: 0.000076  grad: 0.1040 (0.1162)  loss: 0.8160 (0.8102)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [46]  [ 900/6250]  eta: 0:24:47  lr: 0.000076  grad: 0.1067 (0.1157)  loss: 0.8044 (0.8101)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [46]  [1000/6250]  eta: 0:24:16  lr: 0.000076  grad: 0.1022 (0.1152)  loss: 0.8085 (0.8100)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [46]  [1100/6250]  eta: 0:23:44  lr: 0.000075  grad: 0.1091 (0.1149)  loss: 0.8007 (0.8094)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [46]  [1200/6250]  eta: 0:23:14  lr: 0.000075  grad: 0.1100 (0.1149)  loss: 0.8043 (0.8088)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [46]  [1300/6250]  eta: 0:22:44  lr: 0.000075  grad: 0.1085 (0.1150)  loss: 0.8045 (0.8084)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [46]  [1400/6250]  eta: 0:22:30  lr: 0.000075  grad: 0.1084 (0.1151)  loss: 0.8033 (0.8079)  time: 0.3216  data: 0.0435  max mem: 26157
Train: [46]  [1500/6250]  eta: 0:22:00  lr: 0.000075  grad: 0.1087 (0.1155)  loss: 0.7983 (0.8076)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [46]  [1600/6250]  eta: 0:21:30  lr: 0.000075  grad: 0.1091 (0.1156)  loss: 0.8077 (0.8074)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [46]  [1700/6250]  eta: 0:21:00  lr: 0.000075  grad: 0.1062 (0.1161)  loss: 0.8106 (0.8073)  time: 0.2757  data: 0.0002  max mem: 26157
Train: [46]  [1800/6250]  eta: 0:20:31  lr: 0.000075  grad: 0.1077 (0.1162)  loss: 0.8042 (0.8072)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [46]  [1900/6250]  eta: 0:20:01  lr: 0.000075  grad: 0.1116 (0.1161)  loss: 0.8101 (0.8071)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [46]  [2000/6250]  eta: 0:19:41  lr: 0.000075  grad: 0.1149 (0.1160)  loss: 0.8062 (0.8069)  time: 0.4551  data: 0.1782  max mem: 26157
Train: [46]  [2100/6250]  eta: 0:19:17  lr: 0.000075  grad: 0.1159 (0.1163)  loss: 0.8054 (0.8067)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [46]  [2200/6250]  eta: 0:18:48  lr: 0.000075  grad: 0.1123 (0.1164)  loss: 0.8021 (0.8066)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [46]  [2300/6250]  eta: 0:18:19  lr: 0.000075  grad: 0.1082 (0.1166)  loss: 0.8123 (0.8064)  time: 0.2683  data: 0.0003  max mem: 26157
Train: [46]  [2400/6250]  eta: 0:17:49  lr: 0.000075  grad: 0.1067 (0.1170)  loss: 0.8076 (0.8063)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [46]  [2500/6250]  eta: 0:17:20  lr: 0.000075  grad: 0.1048 (0.1172)  loss: 0.8079 (0.8063)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [46]  [2600/6250]  eta: 0:16:51  lr: 0.000075  grad: 0.1131 (0.1171)  loss: 0.8008 (0.8063)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [46]  [2700/6250]  eta: 0:16:23  lr: 0.000075  grad: 0.1151 (0.1171)  loss: 0.8053 (0.8063)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [46]  [2800/6250]  eta: 0:15:54  lr: 0.000075  grad: 0.1090 (0.1171)  loss: 0.8030 (0.8062)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [46]  [2900/6250]  eta: 0:15:26  lr: 0.000075  grad: 0.1062 (0.1170)  loss: 0.8084 (0.8062)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [46]  [3000/6250]  eta: 0:14:57  lr: 0.000075  grad: 0.1091 (0.1171)  loss: 0.8087 (0.8061)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [46]  [3100/6250]  eta: 0:14:29  lr: 0.000075  grad: 0.1028 (0.1169)  loss: 0.8095 (0.8062)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [46]  [3200/6250]  eta: 0:14:01  lr: 0.000075  grad: 0.1067 (0.1168)  loss: 0.8078 (0.8062)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [46]  [3300/6250]  eta: 0:13:33  lr: 0.000075  grad: 0.1053 (0.1167)  loss: 0.8097 (0.8062)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [46]  [3400/6250]  eta: 0:13:05  lr: 0.000075  grad: 0.1084 (0.1166)  loss: 0.8045 (0.8061)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [46]  [3500/6250]  eta: 0:12:40  lr: 0.000075  grad: 0.1146 (0.1166)  loss: 0.7969 (0.8060)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [46]  [3600/6250]  eta: 0:12:12  lr: 0.000075  grad: 0.1115 (0.1165)  loss: 0.8060 (0.8060)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [46]  [3700/6250]  eta: 0:11:46  lr: 0.000075  grad: 0.1075 (0.1166)  loss: 0.8038 (0.8059)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [46]  [3800/6250]  eta: 0:11:18  lr: 0.000075  grad: 0.1255 (0.1168)  loss: 0.8081 (0.8058)  time: 0.2791  data: 0.0002  max mem: 26157
Train: [46]  [3900/6250]  eta: 0:10:51  lr: 0.000075  grad: 0.1104 (0.1169)  loss: 0.7979 (0.8057)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [46]  [4000/6250]  eta: 0:10:23  lr: 0.000075  grad: 0.1072 (0.1168)  loss: 0.8116 (0.8057)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [46]  [4100/6250]  eta: 0:09:55  lr: 0.000075  grad: 0.1124 (0.1169)  loss: 0.7993 (0.8057)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [46]  [4200/6250]  eta: 0:09:27  lr: 0.000074  grad: 0.1110 (0.1169)  loss: 0.8111 (0.8057)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [46]  [4300/6250]  eta: 0:08:59  lr: 0.000074  grad: 0.1171 (0.1170)  loss: 0.7983 (0.8056)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [46]  [4400/6250]  eta: 0:08:31  lr: 0.000074  grad: 0.1097 (0.1170)  loss: 0.8025 (0.8055)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [46]  [4500/6250]  eta: 0:08:06  lr: 0.000074  grad: 0.1244 (0.1172)  loss: 0.8015 (0.8054)  time: 0.7281  data: 0.4553  max mem: 26157
Train: [46]  [4600/6250]  eta: 0:07:38  lr: 0.000074  grad: 0.1210 (0.1173)  loss: 0.7976 (0.8053)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [46]  [4700/6250]  eta: 0:07:11  lr: 0.000074  grad: 0.1178 (0.1175)  loss: 0.8002 (0.8052)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [46]  [4800/6250]  eta: 0:06:44  lr: 0.000074  grad: 0.1133 (0.1176)  loss: 0.8041 (0.8051)  time: 0.3050  data: 0.0337  max mem: 26157
Train: [46]  [4900/6250]  eta: 0:06:16  lr: 0.000074  grad: 0.1158 (0.1179)  loss: 0.8005 (0.8049)  time: 0.2695  data: 0.0001  max mem: 26157
Train: [46]  [5000/6250]  eta: 0:05:48  lr: 0.000074  grad: 0.1148 (0.1179)  loss: 0.7949 (0.8048)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [46]  [5100/6250]  eta: 0:05:19  lr: 0.000074  grad: 0.1184 (0.1181)  loss: 0.8027 (0.8047)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [46]  [5200/6250]  eta: 0:04:51  lr: 0.000074  grad: 0.1201 (0.1182)  loss: 0.7907 (0.8045)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [46]  [5300/6250]  eta: 0:04:24  lr: 0.000074  grad: 0.1110 (0.1184)  loss: 0.8023 (0.8044)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [46]  [5400/6250]  eta: 0:03:57  lr: 0.000074  grad: 0.1176 (0.1185)  loss: 0.7986 (0.8043)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [46]  [5500/6250]  eta: 0:03:29  lr: 0.000074  grad: 0.1189 (0.1187)  loss: 0.8014 (0.8041)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [46]  [5600/6250]  eta: 0:03:01  lr: 0.000074  grad: 0.1219 (0.1189)  loss: 0.7963 (0.8040)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [46]  [5700/6250]  eta: 0:02:33  lr: 0.000074  grad: 0.1223 (0.1190)  loss: 0.7947 (0.8039)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [46]  [5800/6250]  eta: 0:02:05  lr: 0.000074  grad: 0.1151 (0.1191)  loss: 0.8033 (0.8038)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [46]  [5900/6250]  eta: 0:01:37  lr: 0.000074  grad: 0.1030 (0.1191)  loss: 0.8054 (0.8038)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [46]  [6000/6250]  eta: 0:01:09  lr: 0.000074  grad: 0.1134 (0.1191)  loss: 0.7998 (0.8037)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [46]  [6100/6250]  eta: 0:00:41  lr: 0.000074  grad: 0.1193 (0.1191)  loss: 0.8012 (0.8037)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [46]  [6200/6250]  eta: 0:00:13  lr: 0.000074  grad: 0.1143 (0.1192)  loss: 0.7982 (0.8037)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [46]  [6249/6250]  eta: 0:00:00  lr: 0.000074  grad: 0.1150 (0.1192)  loss: 0.8001 (0.8037)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [46] Total time: 0:29:12 (0.2804 s / it)
Averaged stats: lr: 0.000074  grad: 0.1150 (0.1192)  loss: 0.8001 (0.8037)
Eval (hcp-train-subset): [46]  [ 0/62]  eta: 0:04:50  loss: 0.8340 (0.8340)  time: 4.6825  data: 4.5975  max mem: 26157
Eval (hcp-train-subset): [46]  [61/62]  eta: 0:00:00  loss: 0.8229 (0.8251)  time: 0.1299  data: 0.0467  max mem: 26157
Eval (hcp-train-subset): [46] Total time: 0:00:13 (0.2115 s / it)
Averaged stats (hcp-train-subset): loss: 0.8229 (0.8251)
Making plots (hcp-train-subset): example=17
Eval (hcp-val): [46]  [ 0/62]  eta: 0:05:53  loss: 0.8203 (0.8203)  time: 5.7096  data: 5.6253  max mem: 26157
Eval (hcp-val): [46]  [61/62]  eta: 0:00:00  loss: 0.8251 (0.8257)  time: 0.1276  data: 0.0451  max mem: 26157
Eval (hcp-val): [46] Total time: 0:00:13 (0.2172 s / it)
Averaged stats (hcp-val): loss: 0.8251 (0.8257)
Making plots (hcp-val): example=22
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [47]  [   0/6250]  eta: 11:48:05  lr: 0.000074  grad: 0.1680 (0.1680)  loss: 0.8491 (0.8491)  time: 6.7977  data: 6.5155  max mem: 26157
Train: [47]  [ 100/6250]  eta: 0:34:33  lr: 0.000074  grad: 0.1070 (0.1464)  loss: 0.8114 (0.8145)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [47]  [ 200/6250]  eta: 0:30:35  lr: 0.000074  grad: 0.1067 (0.1335)  loss: 0.8087 (0.8105)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [47]  [ 300/6250]  eta: 0:28:58  lr: 0.000074  grad: 0.1160 (0.1294)  loss: 0.7979 (0.8082)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [47]  [ 400/6250]  eta: 0:27:56  lr: 0.000074  grad: 0.1159 (0.1281)  loss: 0.8007 (0.8066)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [47]  [ 500/6250]  eta: 0:27:08  lr: 0.000074  grad: 0.1116 (0.1262)  loss: 0.8000 (0.8061)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [47]  [ 600/6250]  eta: 0:26:27  lr: 0.000074  grad: 0.1107 (0.1254)  loss: 0.8059 (0.8057)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [47]  [ 700/6250]  eta: 0:25:51  lr: 0.000074  grad: 0.1051 (0.1239)  loss: 0.7996 (0.8054)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [47]  [ 800/6250]  eta: 0:25:19  lr: 0.000074  grad: 0.1071 (0.1222)  loss: 0.8064 (0.8053)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [47]  [ 900/6250]  eta: 0:24:54  lr: 0.000074  grad: 0.1110 (0.1222)  loss: 0.8060 (0.8053)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [47]  [1000/6250]  eta: 0:24:36  lr: 0.000073  grad: 0.1116 (0.1211)  loss: 0.8008 (0.8054)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [47]  [1100/6250]  eta: 0:24:07  lr: 0.000073  grad: 0.1104 (0.1207)  loss: 0.8005 (0.8053)  time: 0.2707  data: 0.0001  max mem: 26157
Train: [47]  [1200/6250]  eta: 0:23:34  lr: 0.000073  grad: 0.1133 (0.1204)  loss: 0.8034 (0.8051)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [47]  [1300/6250]  eta: 0:23:06  lr: 0.000073  grad: 0.1196 (0.1204)  loss: 0.8010 (0.8050)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [47]  [1400/6250]  eta: 0:22:35  lr: 0.000073  grad: 0.1043 (0.1201)  loss: 0.8058 (0.8049)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [47]  [1500/6250]  eta: 0:22:03  lr: 0.000073  grad: 0.1151 (0.1201)  loss: 0.8017 (0.8048)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [47]  [1600/6250]  eta: 0:21:33  lr: 0.000073  grad: 0.1184 (0.1196)  loss: 0.7980 (0.8047)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [47]  [1700/6250]  eta: 0:21:03  lr: 0.000073  grad: 0.1147 (0.1196)  loss: 0.7999 (0.8047)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [47]  [1800/6250]  eta: 0:20:34  lr: 0.000073  grad: 0.1098 (0.1198)  loss: 0.8053 (0.8047)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [47]  [1900/6250]  eta: 0:20:04  lr: 0.000073  grad: 0.1140 (0.1198)  loss: 0.7985 (0.8045)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [47]  [2000/6250]  eta: 0:19:35  lr: 0.000073  grad: 0.1140 (0.1201)  loss: 0.8022 (0.8044)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [47]  [2100/6250]  eta: 0:19:06  lr: 0.000073  grad: 0.1170 (0.1202)  loss: 0.7980 (0.8043)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [47]  [2200/6250]  eta: 0:18:37  lr: 0.000073  grad: 0.1069 (0.1199)  loss: 0.8031 (0.8041)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [47]  [2300/6250]  eta: 0:18:08  lr: 0.000073  grad: 0.1072 (0.1197)  loss: 0.7999 (0.8040)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [47]  [2400/6250]  eta: 0:17:40  lr: 0.000073  grad: 0.1153 (0.1195)  loss: 0.8049 (0.8040)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [47]  [2500/6250]  eta: 0:17:11  lr: 0.000073  grad: 0.1131 (0.1197)  loss: 0.8022 (0.8039)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [47]  [2600/6250]  eta: 0:16:43  lr: 0.000073  grad: 0.1052 (0.1196)  loss: 0.8098 (0.8039)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [47]  [2700/6250]  eta: 0:16:23  lr: 0.000073  grad: 0.1119 (0.1196)  loss: 0.8082 (0.8040)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [47]  [2800/6250]  eta: 0:15:55  lr: 0.000073  grad: 0.1119 (0.1198)  loss: 0.8053 (0.8040)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [47]  [2900/6250]  eta: 0:15:36  lr: 0.000073  grad: 0.1169 (0.1199)  loss: 0.8032 (0.8040)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [47]  [3000/6250]  eta: 0:15:07  lr: 0.000073  grad: 0.1168 (0.1200)  loss: 0.7962 (0.8039)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [47]  [3100/6250]  eta: 0:14:40  lr: 0.000073  grad: 0.1143 (0.1200)  loss: 0.7980 (0.8039)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [47]  [3200/6250]  eta: 0:14:12  lr: 0.000073  grad: 0.1187 (0.1202)  loss: 0.7979 (0.8038)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [47]  [3300/6250]  eta: 0:13:43  lr: 0.000073  grad: 0.1179 (0.1203)  loss: 0.7942 (0.8037)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [47]  [3400/6250]  eta: 0:13:15  lr: 0.000073  grad: 0.1165 (0.1204)  loss: 0.8002 (0.8035)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [47]  [3500/6250]  eta: 0:12:46  lr: 0.000073  grad: 0.1131 (0.1206)  loss: 0.8016 (0.8034)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [47]  [3600/6250]  eta: 0:12:18  lr: 0.000073  grad: 0.1206 (0.1208)  loss: 0.7981 (0.8033)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [47]  [3700/6250]  eta: 0:11:49  lr: 0.000073  grad: 0.1222 (0.1211)  loss: 0.8026 (0.8032)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [47]  [3800/6250]  eta: 0:11:21  lr: 0.000073  grad: 0.1192 (0.1214)  loss: 0.7998 (0.8030)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [47]  [3900/6250]  eta: 0:10:52  lr: 0.000073  grad: 0.1218 (0.1217)  loss: 0.7996 (0.8029)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [47]  [4000/6250]  eta: 0:10:24  lr: 0.000073  grad: 0.1167 (0.1218)  loss: 0.7941 (0.8028)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [47]  [4100/6250]  eta: 0:09:56  lr: 0.000072  grad: 0.1296 (0.1221)  loss: 0.7934 (0.8026)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [47]  [4200/6250]  eta: 0:09:28  lr: 0.000072  grad: 0.1144 (0.1222)  loss: 0.7925 (0.8025)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [47]  [4300/6250]  eta: 0:09:00  lr: 0.000072  grad: 0.1268 (0.1223)  loss: 0.7948 (0.8023)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [47]  [4400/6250]  eta: 0:08:32  lr: 0.000072  grad: 0.1210 (0.1225)  loss: 0.7977 (0.8022)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [47]  [4500/6250]  eta: 0:08:04  lr: 0.000072  grad: 0.1346 (0.1228)  loss: 0.7913 (0.8021)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [47]  [4600/6250]  eta: 0:07:36  lr: 0.000072  grad: 0.1175 (0.1230)  loss: 0.8010 (0.8020)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [47]  [4700/6250]  eta: 0:07:08  lr: 0.000072  grad: 0.1213 (0.1231)  loss: 0.8043 (0.8019)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [47]  [4800/6250]  eta: 0:06:40  lr: 0.000072  grad: 0.1200 (0.1231)  loss: 0.8036 (0.8019)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [47]  [4900/6250]  eta: 0:06:12  lr: 0.000072  grad: 0.1151 (0.1232)  loss: 0.8000 (0.8019)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [47]  [5000/6250]  eta: 0:05:45  lr: 0.000072  grad: 0.1141 (0.1232)  loss: 0.8023 (0.8019)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [47]  [5100/6250]  eta: 0:05:17  lr: 0.000072  grad: 0.1281 (0.1233)  loss: 0.8007 (0.8019)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [47]  [5200/6250]  eta: 0:04:49  lr: 0.000072  grad: 0.1175 (0.1233)  loss: 0.8062 (0.8019)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [47]  [5300/6250]  eta: 0:04:22  lr: 0.000072  grad: 0.1233 (0.1233)  loss: 0.8011 (0.8019)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [47]  [5400/6250]  eta: 0:03:54  lr: 0.000072  grad: 0.1260 (0.1234)  loss: 0.8084 (0.8019)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [47]  [5500/6250]  eta: 0:03:26  lr: 0.000072  grad: 0.1146 (0.1235)  loss: 0.8068 (0.8019)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [47]  [5600/6250]  eta: 0:02:59  lr: 0.000072  grad: 0.1218 (0.1236)  loss: 0.8038 (0.8019)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [47]  [5700/6250]  eta: 0:02:31  lr: 0.000072  grad: 0.1421 (0.1238)  loss: 0.8003 (0.8019)  time: 0.2706  data: 0.0001  max mem: 26157
Train: [47]  [5800/6250]  eta: 0:02:03  lr: 0.000072  grad: 0.1229 (0.1240)  loss: 0.8001 (0.8020)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [47]  [5900/6250]  eta: 0:01:36  lr: 0.000072  grad: 0.1288 (0.1243)  loss: 0.8003 (0.8019)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [47]  [6000/6250]  eta: 0:01:08  lr: 0.000072  grad: 0.1288 (0.1244)  loss: 0.8041 (0.8020)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [47]  [6100/6250]  eta: 0:00:41  lr: 0.000072  grad: 0.1198 (0.1245)  loss: 0.7982 (0.8020)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [47]  [6200/6250]  eta: 0:00:13  lr: 0.000072  grad: 0.1109 (0.1247)  loss: 0.8023 (0.8020)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [47]  [6249/6250]  eta: 0:00:00  lr: 0.000072  grad: 0.1141 (0.1247)  loss: 0.7994 (0.8020)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [47] Total time: 0:28:44 (0.2759 s / it)
Averaged stats: lr: 0.000072  grad: 0.1141 (0.1247)  loss: 0.7994 (0.8020)
Eval (hcp-train-subset): [47]  [ 0/62]  eta: 0:04:07  loss: 0.8321 (0.8321)  time: 3.9885  data: 3.8596  max mem: 26157
Eval (hcp-train-subset): [47]  [61/62]  eta: 0:00:00  loss: 0.8242 (0.8252)  time: 0.1483  data: 0.0639  max mem: 26157
Eval (hcp-train-subset): [47] Total time: 0:00:13 (0.2213 s / it)
Averaged stats (hcp-train-subset): loss: 0.8242 (0.8252)
Making plots (hcp-train-subset): example=49
Eval (hcp-val): [47]  [ 0/62]  eta: 0:03:56  loss: 0.8231 (0.8231)  time: 3.8209  data: 3.7141  max mem: 26157
Eval (hcp-val): [47]  [61/62]  eta: 0:00:00  loss: 0.8249 (0.8257)  time: 0.1304  data: 0.0475  max mem: 26157
Eval (hcp-val): [47] Total time: 0:00:13 (0.2121 s / it)
Averaged stats (hcp-val): loss: 0.8249 (0.8257)
Making plots (hcp-val): example=28
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [48]  [   0/6250]  eta: 11:24:22  lr: 0.000072  grad: 0.3552 (0.3552)  loss: 0.8059 (0.8059)  time: 6.5699  data: 6.2852  max mem: 26157
Train: [48]  [ 100/6250]  eta: 0:34:08  lr: 0.000072  grad: 0.1140 (0.1467)  loss: 0.8172 (0.8152)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [48]  [ 200/6250]  eta: 0:30:22  lr: 0.000072  grad: 0.1234 (0.1360)  loss: 0.7981 (0.8104)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [48]  [ 300/6250]  eta: 0:28:53  lr: 0.000072  grad: 0.1242 (0.1334)  loss: 0.7985 (0.8075)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [48]  [ 400/6250]  eta: 0:27:52  lr: 0.000072  grad: 0.1189 (0.1324)  loss: 0.7937 (0.8048)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [48]  [ 500/6250]  eta: 0:27:04  lr: 0.000072  grad: 0.1126 (0.1318)  loss: 0.8052 (0.8039)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [48]  [ 600/6250]  eta: 0:26:25  lr: 0.000072  grad: 0.1083 (0.1317)  loss: 0.7898 (0.8026)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [48]  [ 700/6250]  eta: 0:25:50  lr: 0.000072  grad: 0.1104 (0.1304)  loss: 0.8038 (0.8022)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [48]  [ 800/6250]  eta: 0:25:16  lr: 0.000072  grad: 0.1072 (0.1292)  loss: 0.7950 (0.8018)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [48]  [ 900/6250]  eta: 0:25:38  lr: 0.000071  grad: 0.1173 (0.1277)  loss: 0.8038 (0.8015)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [48]  [1000/6250]  eta: 0:25:12  lr: 0.000071  grad: 0.1251 (0.1281)  loss: 0.8026 (0.8012)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [48]  [1100/6250]  eta: 0:24:35  lr: 0.000071  grad: 0.1160 (0.1283)  loss: 0.8017 (0.8011)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [48]  [1200/6250]  eta: 0:24:00  lr: 0.000071  grad: 0.1133 (0.1278)  loss: 0.7895 (0.8008)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [48]  [1300/6250]  eta: 0:23:25  lr: 0.000071  grad: 0.1128 (0.1272)  loss: 0.7999 (0.8006)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [48]  [1400/6250]  eta: 0:22:52  lr: 0.000071  grad: 0.1117 (0.1268)  loss: 0.7951 (0.8004)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [48]  [1500/6250]  eta: 0:22:19  lr: 0.000071  grad: 0.1183 (0.1266)  loss: 0.8026 (0.8004)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [48]  [1600/6250]  eta: 0:21:48  lr: 0.000071  grad: 0.1139 (0.1262)  loss: 0.7983 (0.8003)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [48]  [1700/6250]  eta: 0:21:16  lr: 0.000071  grad: 0.1217 (0.1263)  loss: 0.8035 (0.8002)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [48]  [1800/6250]  eta: 0:20:45  lr: 0.000071  grad: 0.1177 (0.1267)  loss: 0.7926 (0.8002)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [48]  [1900/6250]  eta: 0:20:15  lr: 0.000071  grad: 0.1226 (0.1264)  loss: 0.8071 (0.8002)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [48]  [2000/6250]  eta: 0:19:45  lr: 0.000071  grad: 0.1191 (0.1264)  loss: 0.8015 (0.8001)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [48]  [2100/6250]  eta: 0:19:16  lr: 0.000071  grad: 0.1270 (0.1265)  loss: 0.8013 (0.8001)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [48]  [2200/6250]  eta: 0:18:46  lr: 0.000071  grad: 0.1199 (0.1264)  loss: 0.7970 (0.8001)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [48]  [2300/6250]  eta: 0:18:17  lr: 0.000071  grad: 0.1146 (0.1264)  loss: 0.7995 (0.8000)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [48]  [2400/6250]  eta: 0:17:50  lr: 0.000071  grad: 0.1196 (0.1265)  loss: 0.8017 (0.7999)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [48]  [2500/6250]  eta: 0:17:22  lr: 0.000071  grad: 0.1311 (0.1267)  loss: 0.8026 (0.7998)  time: 0.2741  data: 0.0002  max mem: 26157
Train: [48]  [2600/6250]  eta: 0:16:53  lr: 0.000071  grad: 0.1224 (0.1268)  loss: 0.7961 (0.7997)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [48]  [2700/6250]  eta: 0:16:25  lr: 0.000071  grad: 0.1197 (0.1268)  loss: 0.8063 (0.7997)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [48]  [2800/6250]  eta: 0:15:56  lr: 0.000071  grad: 0.1177 (0.1269)  loss: 0.8037 (0.7998)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [48]  [2900/6250]  eta: 0:15:28  lr: 0.000071  grad: 0.1184 (0.1269)  loss: 0.8018 (0.7997)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [48]  [3000/6250]  eta: 0:14:59  lr: 0.000071  grad: 0.1203 (0.1267)  loss: 0.8047 (0.7999)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [48]  [3100/6250]  eta: 0:14:31  lr: 0.000071  grad: 0.1188 (0.1267)  loss: 0.8018 (0.7999)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [48]  [3200/6250]  eta: 0:14:02  lr: 0.000071  grad: 0.1170 (0.1268)  loss: 0.8039 (0.8000)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [48]  [3300/6250]  eta: 0:13:34  lr: 0.000071  grad: 0.1227 (0.1268)  loss: 0.7974 (0.8000)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [48]  [3400/6250]  eta: 0:13:06  lr: 0.000071  grad: 0.1175 (0.1267)  loss: 0.8033 (0.8000)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [48]  [3500/6250]  eta: 0:12:38  lr: 0.000071  grad: 0.1247 (0.1265)  loss: 0.8045 (0.8000)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [48]  [3600/6250]  eta: 0:12:10  lr: 0.000071  grad: 0.1130 (0.1265)  loss: 0.8000 (0.8000)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [48]  [3700/6250]  eta: 0:11:42  lr: 0.000071  grad: 0.1176 (0.1264)  loss: 0.7928 (0.7999)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [48]  [3800/6250]  eta: 0:11:14  lr: 0.000071  grad: 0.1204 (0.1264)  loss: 0.7982 (0.7999)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [48]  [3900/6250]  eta: 0:10:46  lr: 0.000070  grad: 0.1349 (0.1264)  loss: 0.7979 (0.7998)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [48]  [4000/6250]  eta: 0:10:18  lr: 0.000070  grad: 0.1147 (0.1264)  loss: 0.7987 (0.7998)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [48]  [4100/6250]  eta: 0:09:50  lr: 0.000070  grad: 0.1302 (0.1264)  loss: 0.7876 (0.7997)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [48]  [4200/6250]  eta: 0:09:22  lr: 0.000070  grad: 0.1154 (0.1264)  loss: 0.8053 (0.7997)  time: 0.2694  data: 0.0001  max mem: 26157
Train: [48]  [4300/6250]  eta: 0:08:55  lr: 0.000070  grad: 0.1153 (0.1263)  loss: 0.7943 (0.7998)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [48]  [4400/6250]  eta: 0:08:30  lr: 0.000070  grad: 0.1186 (0.1262)  loss: 0.8045 (0.7998)  time: 0.6181  data: 0.3437  max mem: 26157
Train: [48]  [4500/6250]  eta: 0:08:02  lr: 0.000070  grad: 0.1235 (0.1262)  loss: 0.7974 (0.7998)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [48]  [4600/6250]  eta: 0:07:34  lr: 0.000070  grad: 0.1173 (0.1261)  loss: 0.8003 (0.7998)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [48]  [4700/6250]  eta: 0:07:07  lr: 0.000070  grad: 0.1165 (0.1260)  loss: 0.7981 (0.7999)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [48]  [4800/6250]  eta: 0:06:39  lr: 0.000070  grad: 0.1177 (0.1260)  loss: 0.8041 (0.7999)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [48]  [4900/6250]  eta: 0:06:11  lr: 0.000070  grad: 0.1121 (0.1259)  loss: 0.7977 (0.7999)  time: 0.2693  data: 0.0004  max mem: 26157
Train: [48]  [5000/6250]  eta: 0:05:44  lr: 0.000070  grad: 0.1316 (0.1259)  loss: 0.7959 (0.7999)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [48]  [5100/6250]  eta: 0:05:16  lr: 0.000070  grad: 0.1171 (0.1258)  loss: 0.8040 (0.7999)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [48]  [5200/6250]  eta: 0:04:48  lr: 0.000070  grad: 0.1170 (0.1257)  loss: 0.7946 (0.7999)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [48]  [5300/6250]  eta: 0:04:21  lr: 0.000070  grad: 0.1211 (0.1257)  loss: 0.8005 (0.8000)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [48]  [5400/6250]  eta: 0:03:53  lr: 0.000070  grad: 0.1177 (0.1257)  loss: 0.8046 (0.8000)  time: 0.2704  data: 0.0001  max mem: 26157
Train: [48]  [5500/6250]  eta: 0:03:26  lr: 0.000070  grad: 0.1198 (0.1257)  loss: 0.7986 (0.8000)  time: 0.2695  data: 0.0001  max mem: 26157
Train: [48]  [5600/6250]  eta: 0:02:58  lr: 0.000070  grad: 0.1207 (0.1257)  loss: 0.7991 (0.8000)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [48]  [5700/6250]  eta: 0:02:31  lr: 0.000070  grad: 0.1182 (0.1258)  loss: 0.7942 (0.8000)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [48]  [5800/6250]  eta: 0:02:03  lr: 0.000070  grad: 0.1233 (0.1258)  loss: 0.8021 (0.8000)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [48]  [5900/6250]  eta: 0:01:36  lr: 0.000070  grad: 0.1131 (0.1258)  loss: 0.8072 (0.8001)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [48]  [6000/6250]  eta: 0:01:08  lr: 0.000070  grad: 0.1183 (0.1259)  loss: 0.8037 (0.8002)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [48]  [6100/6250]  eta: 0:00:41  lr: 0.000070  grad: 0.1219 (0.1258)  loss: 0.8028 (0.8002)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [48]  [6200/6250]  eta: 0:00:13  lr: 0.000070  grad: 0.1128 (0.1258)  loss: 0.8050 (0.8003)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [48]  [6249/6250]  eta: 0:00:00  lr: 0.000070  grad: 0.1211 (0.1258)  loss: 0.8038 (0.8003)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [48] Total time: 0:28:40 (0.2754 s / it)
Averaged stats: lr: 0.000070  grad: 0.1211 (0.1258)  loss: 0.8038 (0.8003)
Eval (hcp-train-subset): [48]  [ 0/62]  eta: 0:05:43  loss: 0.8359 (0.8359)  time: 5.5469  data: 5.4625  max mem: 26157
Eval (hcp-train-subset): [48]  [61/62]  eta: 0:00:00  loss: 0.8239 (0.8236)  time: 0.1148  data: 0.0320  max mem: 26157
Eval (hcp-train-subset): [48] Total time: 0:00:13 (0.2119 s / it)
Averaged stats (hcp-train-subset): loss: 0.8239 (0.8236)
Making plots (hcp-train-subset): example=23
Eval (hcp-val): [48]  [ 0/62]  eta: 0:04:14  loss: 0.8208 (0.8208)  time: 4.0992  data: 3.9946  max mem: 26157
Eval (hcp-val): [48]  [61/62]  eta: 0:00:00  loss: 0.8251 (0.8260)  time: 0.1293  data: 0.0448  max mem: 26157
Eval (hcp-val): [48] Total time: 0:00:13 (0.2195 s / it)
Averaged stats (hcp-val): loss: 0.8251 (0.8260)
Making plots (hcp-val): example=50
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [49]  [   0/6250]  eta: 8:18:45  lr: 0.000070  grad: 0.1242 (0.1242)  loss: 0.8461 (0.8461)  time: 4.7881  data: 4.4646  max mem: 26157
Train: [49]  [ 100/6250]  eta: 0:34:02  lr: 0.000070  grad: 0.1253 (0.1704)  loss: 0.8040 (0.8074)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [49]  [ 200/6250]  eta: 0:30:27  lr: 0.000070  grad: 0.1308 (0.1512)  loss: 0.7955 (0.8047)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [49]  [ 300/6250]  eta: 0:28:57  lr: 0.000070  grad: 0.1114 (0.1421)  loss: 0.8100 (0.8057)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [49]  [ 400/6250]  eta: 0:27:57  lr: 0.000070  grad: 0.1104 (0.1357)  loss: 0.8069 (0.8063)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [49]  [ 500/6250]  eta: 0:27:09  lr: 0.000070  grad: 0.1217 (0.1312)  loss: 0.8024 (0.8064)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [49]  [ 600/6250]  eta: 0:26:27  lr: 0.000070  grad: 0.1107 (0.1297)  loss: 0.8111 (0.8068)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [49]  [ 700/6250]  eta: 0:25:50  lr: 0.000069  grad: 0.1051 (0.1272)  loss: 0.8080 (0.8071)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [49]  [ 800/6250]  eta: 0:26:25  lr: 0.000069  grad: 0.1111 (0.1261)  loss: 0.8081 (0.8069)  time: 0.2792  data: 0.0004  max mem: 26157
Train: [49]  [ 900/6250]  eta: 0:26:36  lr: 0.000069  grad: 0.1205 (0.1258)  loss: 0.8089 (0.8068)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [49]  [1000/6250]  eta: 0:26:07  lr: 0.000069  grad: 0.1111 (0.1250)  loss: 0.8064 (0.8068)  time: 0.2767  data: 0.0002  max mem: 26157
Train: [49]  [1100/6250]  eta: 0:26:41  lr: 0.000069  grad: 0.1068 (0.1243)  loss: 0.8020 (0.8066)  time: 0.6763  data: 0.3973  max mem: 26157
Train: [49]  [1200/6250]  eta: 0:25:54  lr: 0.000069  grad: 0.1137 (0.1239)  loss: 0.8026 (0.8062)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [49]  [1300/6250]  eta: 0:25:09  lr: 0.000069  grad: 0.1161 (0.1235)  loss: 0.8070 (0.8058)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [49]  [1400/6250]  eta: 0:25:03  lr: 0.000069  grad: 0.1120 (0.1231)  loss: 0.7975 (0.8055)  time: 0.7630  data: 0.4877  max mem: 26157
Train: [49]  [1500/6250]  eta: 0:24:47  lr: 0.000069  grad: 0.1181 (0.1227)  loss: 0.7961 (0.8050)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [49]  [1600/6250]  eta: 0:24:04  lr: 0.000069  grad: 0.1270 (0.1229)  loss: 0.7928 (0.8044)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [49]  [1700/6250]  eta: 0:23:22  lr: 0.000069  grad: 0.1230 (0.1231)  loss: 0.8013 (0.8041)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [49]  [1800/6250]  eta: 0:22:41  lr: 0.000069  grad: 0.1128 (0.1233)  loss: 0.7981 (0.8038)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [49]  [1900/6250]  eta: 0:22:17  lr: 0.000069  grad: 0.1160 (0.1236)  loss: 0.7888 (0.8034)  time: 0.2817  data: 0.0003  max mem: 26157
Train: [49]  [2000/6250]  eta: 0:21:46  lr: 0.000069  grad: 0.1195 (0.1237)  loss: 0.7943 (0.8030)  time: 0.2719  data: 0.0004  max mem: 26157
Train: [49]  [2100/6250]  eta: 0:21:08  lr: 0.000069  grad: 0.1238 (0.1239)  loss: 0.7948 (0.8027)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [49]  [2200/6250]  eta: 0:20:31  lr: 0.000069  grad: 0.1096 (0.1241)  loss: 0.8012 (0.8025)  time: 0.2750  data: 0.0003  max mem: 26157
Train: [49]  [2300/6250]  eta: 0:19:54  lr: 0.000069  grad: 0.1169 (0.1242)  loss: 0.7982 (0.8022)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [49]  [2400/6250]  eta: 0:19:19  lr: 0.000069  grad: 0.1216 (0.1243)  loss: 0.8028 (0.8022)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [49]  [2500/6250]  eta: 0:18:44  lr: 0.000069  grad: 0.1277 (0.1245)  loss: 0.7942 (0.8019)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [49]  [2600/6250]  eta: 0:18:10  lr: 0.000069  grad: 0.1247 (0.1249)  loss: 0.7879 (0.8016)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [49]  [2700/6250]  eta: 0:18:01  lr: 0.000069  grad: 0.1156 (0.1251)  loss: 0.7925 (0.8014)  time: 0.6583  data: 0.3776  max mem: 26157
Train: [49]  [2800/6250]  eta: 0:17:33  lr: 0.000069  grad: 0.1174 (0.1255)  loss: 0.7941 (0.8012)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [49]  [2900/6250]  eta: 0:16:59  lr: 0.000069  grad: 0.1260 (0.1258)  loss: 0.7931 (0.8010)  time: 0.2889  data: 0.0090  max mem: 26157
Train: [49]  [3000/6250]  eta: 0:16:39  lr: 0.000069  grad: 0.1193 (0.1258)  loss: 0.8025 (0.8009)  time: 0.3992  data: 0.1276  max mem: 26157
Train: [49]  [3100/6250]  eta: 0:16:05  lr: 0.000069  grad: 0.1236 (0.1261)  loss: 0.8009 (0.8008)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [49]  [3200/6250]  eta: 0:15:41  lr: 0.000069  grad: 0.1287 (0.1264)  loss: 0.8021 (0.8007)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [49]  [3300/6250]  eta: 0:15:07  lr: 0.000069  grad: 0.1208 (0.1266)  loss: 0.7917 (0.8005)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [49]  [3400/6250]  eta: 0:14:33  lr: 0.000069  grad: 0.1287 (0.1267)  loss: 0.7890 (0.8003)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [49]  [3500/6250]  eta: 0:13:59  lr: 0.000069  grad: 0.1192 (0.1269)  loss: 0.7935 (0.8001)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [49]  [3600/6250]  eta: 0:13:26  lr: 0.000069  grad: 0.1416 (0.1274)  loss: 0.7943 (0.7999)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [49]  [3700/6250]  eta: 0:12:53  lr: 0.000069  grad: 0.1255 (0.1278)  loss: 0.7892 (0.7997)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [49]  [3800/6250]  eta: 0:12:21  lr: 0.000068  grad: 0.1330 (0.1281)  loss: 0.7953 (0.7996)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [49]  [3900/6250]  eta: 0:11:49  lr: 0.000068  grad: 0.1206 (0.1283)  loss: 0.7943 (0.7995)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [49]  [4000/6250]  eta: 0:11:17  lr: 0.000068  grad: 0.1169 (0.1282)  loss: 0.7904 (0.7994)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [49]  [4100/6250]  eta: 0:10:45  lr: 0.000068  grad: 0.1127 (0.1283)  loss: 0.7982 (0.7992)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [49]  [4200/6250]  eta: 0:10:14  lr: 0.000068  grad: 0.1267 (0.1283)  loss: 0.8030 (0.7993)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [49]  [4300/6250]  eta: 0:09:42  lr: 0.000068  grad: 0.1142 (0.1283)  loss: 0.8002 (0.7993)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [49]  [4400/6250]  eta: 0:09:11  lr: 0.000068  grad: 0.1198 (0.1283)  loss: 0.7982 (0.7993)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [49]  [4500/6250]  eta: 0:08:43  lr: 0.000068  grad: 0.1159 (0.1283)  loss: 0.8025 (0.7993)  time: 0.5447  data: 0.2556  max mem: 26157
Train: [49]  [4600/6250]  eta: 0:08:12  lr: 0.000068  grad: 0.1255 (0.1285)  loss: 0.7965 (0.7994)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [49]  [4700/6250]  eta: 0:07:41  lr: 0.000068  grad: 0.1190 (0.1286)  loss: 0.8020 (0.7994)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [49]  [4800/6250]  eta: 0:07:11  lr: 0.000068  grad: 0.1242 (0.1286)  loss: 0.7983 (0.7995)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [49]  [4900/6250]  eta: 0:06:40  lr: 0.000068  grad: 0.1208 (0.1286)  loss: 0.8040 (0.7995)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [49]  [5000/6250]  eta: 0:06:10  lr: 0.000068  grad: 0.1243 (0.1286)  loss: 0.7987 (0.7996)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [49]  [5100/6250]  eta: 0:05:40  lr: 0.000068  grad: 0.1133 (0.1286)  loss: 0.8022 (0.7997)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [49]  [5200/6250]  eta: 0:05:10  lr: 0.000068  grad: 0.1264 (0.1286)  loss: 0.7975 (0.7997)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [49]  [5300/6250]  eta: 0:04:40  lr: 0.000068  grad: 0.1145 (0.1286)  loss: 0.8038 (0.7997)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [49]  [5400/6250]  eta: 0:04:10  lr: 0.000068  grad: 0.1269 (0.1287)  loss: 0.8060 (0.7997)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [49]  [5500/6250]  eta: 0:03:40  lr: 0.000068  grad: 0.1165 (0.1288)  loss: 0.8010 (0.7997)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [49]  [5600/6250]  eta: 0:03:10  lr: 0.000068  grad: 0.1253 (0.1289)  loss: 0.7987 (0.7998)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [49]  [5700/6250]  eta: 0:02:41  lr: 0.000068  grad: 0.1166 (0.1288)  loss: 0.8035 (0.7998)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [49]  [5800/6250]  eta: 0:02:11  lr: 0.000068  grad: 0.1244 (0.1288)  loss: 0.7977 (0.7999)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [49]  [5900/6250]  eta: 0:01:42  lr: 0.000068  grad: 0.1167 (0.1288)  loss: 0.8028 (0.7999)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [49]  [6000/6250]  eta: 0:01:12  lr: 0.000068  grad: 0.1104 (0.1289)  loss: 0.8050 (0.8000)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [49]  [6100/6250]  eta: 0:00:43  lr: 0.000068  grad: 0.1213 (0.1289)  loss: 0.7958 (0.8000)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [49]  [6200/6250]  eta: 0:00:14  lr: 0.000068  grad: 0.1239 (0.1290)  loss: 0.8036 (0.8001)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [49]  [6249/6250]  eta: 0:00:00  lr: 0.000068  grad: 0.1254 (0.1290)  loss: 0.7952 (0.8001)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [49] Total time: 0:30:26 (0.2922 s / it)
Averaged stats: lr: 0.000068  grad: 0.1254 (0.1290)  loss: 0.7952 (0.8001)
Eval (hcp-train-subset): [49]  [ 0/62]  eta: 0:04:14  loss: 0.8335 (0.8335)  time: 4.0996  data: 3.9696  max mem: 26157
Eval (hcp-train-subset): [49]  [61/62]  eta: 0:00:00  loss: 0.8240 (0.8249)  time: 0.1281  data: 0.0437  max mem: 26157
Eval (hcp-train-subset): [49] Total time: 0:00:13 (0.2139 s / it)
Averaged stats (hcp-train-subset): loss: 0.8240 (0.8249)
Making plots (hcp-train-subset): example=59
Eval (hcp-val): [49]  [ 0/62]  eta: 0:05:37  loss: 0.8228 (0.8228)  time: 5.4385  data: 5.3537  max mem: 26157
Eval (hcp-val): [49]  [61/62]  eta: 0:00:00  loss: 0.8244 (0.8258)  time: 0.1360  data: 0.0529  max mem: 26157
Eval (hcp-val): [49] Total time: 0:00:13 (0.2211 s / it)
Averaged stats (hcp-val): loss: 0.8244 (0.8258)
Making plots (hcp-val): example=33
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [50]  [   0/6250]  eta: 7:45:08  lr: 0.000068  grad: 0.0737 (0.0737)  loss: 0.8313 (0.8313)  time: 4.4654  data: 4.0874  max mem: 26157
Train: [50]  [ 100/6250]  eta: 0:34:21  lr: 0.000068  grad: 0.1583 (0.1611)  loss: 0.7894 (0.8068)  time: 0.2706  data: 0.0004  max mem: 26157
Train: [50]  [ 200/6250]  eta: 0:30:37  lr: 0.000068  grad: 0.1404 (0.1539)  loss: 0.7894 (0.8019)  time: 0.2732  data: 0.0003  max mem: 26157
Train: [50]  [ 300/6250]  eta: 0:29:00  lr: 0.000068  grad: 0.1279 (0.1510)  loss: 0.8043 (0.8008)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [50]  [ 400/6250]  eta: 0:27:59  lr: 0.000068  grad: 0.1186 (0.1459)  loss: 0.8006 (0.8010)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [50]  [ 500/6250]  eta: 0:27:12  lr: 0.000067  grad: 0.1346 (0.1447)  loss: 0.7976 (0.8009)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [50]  [ 600/6250]  eta: 0:26:30  lr: 0.000067  grad: 0.1195 (0.1446)  loss: 0.8046 (0.8004)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [50]  [ 700/6250]  eta: 0:25:52  lr: 0.000067  grad: 0.1164 (0.1431)  loss: 0.8018 (0.8000)  time: 0.2704  data: 0.0001  max mem: 26157
Train: [50]  [ 800/6250]  eta: 0:25:17  lr: 0.000067  grad: 0.1206 (0.1405)  loss: 0.8104 (0.8003)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [50]  [ 900/6250]  eta: 0:24:45  lr: 0.000067  grad: 0.1194 (0.1407)  loss: 0.8073 (0.8004)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [50]  [1000/6250]  eta: 0:24:13  lr: 0.000067  grad: 0.1180 (0.1397)  loss: 0.8014 (0.8007)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [50]  [1100/6250]  eta: 0:23:42  lr: 0.000067  grad: 0.1110 (0.1377)  loss: 0.8016 (0.8008)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [50]  [1200/6250]  eta: 0:23:12  lr: 0.000067  grad: 0.1110 (0.1372)  loss: 0.7932 (0.8007)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [50]  [1300/6250]  eta: 0:22:42  lr: 0.000067  grad: 0.1115 (0.1363)  loss: 0.8041 (0.8008)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [50]  [1400/6250]  eta: 0:22:13  lr: 0.000067  grad: 0.1190 (0.1361)  loss: 0.7985 (0.8009)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [50]  [1500/6250]  eta: 0:21:44  lr: 0.000067  grad: 0.1146 (0.1355)  loss: 0.8013 (0.8008)  time: 0.2720  data: 0.0004  max mem: 26157
Train: [50]  [1600/6250]  eta: 0:21:25  lr: 0.000067  grad: 0.1043 (0.1345)  loss: 0.8025 (0.8008)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [50]  [1700/6250]  eta: 0:20:56  lr: 0.000067  grad: 0.1159 (0.1342)  loss: 0.8011 (0.8007)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [50]  [1800/6250]  eta: 0:20:27  lr: 0.000067  grad: 0.1153 (0.1339)  loss: 0.8011 (0.8006)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [50]  [1900/6250]  eta: 0:20:16  lr: 0.000067  grad: 0.1175 (0.1339)  loss: 0.8031 (0.8006)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [50]  [2000/6250]  eta: 0:19:46  lr: 0.000067  grad: 0.1334 (0.1341)  loss: 0.7995 (0.8005)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [50]  [2100/6250]  eta: 0:19:16  lr: 0.000067  grad: 0.1181 (0.1339)  loss: 0.7986 (0.8003)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [50]  [2200/6250]  eta: 0:18:47  lr: 0.000067  grad: 0.1186 (0.1336)  loss: 0.7964 (0.8002)  time: 0.2749  data: 0.0002  max mem: 26157
Train: [50]  [2300/6250]  eta: 0:18:25  lr: 0.000067  grad: 0.1203 (0.1337)  loss: 0.7998 (0.7999)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [50]  [2400/6250]  eta: 0:17:55  lr: 0.000067  grad: 0.1184 (0.1332)  loss: 0.7948 (0.7998)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [50]  [2500/6250]  eta: 0:17:26  lr: 0.000067  grad: 0.1132 (0.1328)  loss: 0.7974 (0.7999)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [50]  [2600/6250]  eta: 0:16:56  lr: 0.000067  grad: 0.1201 (0.1326)  loss: 0.7987 (0.7999)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [50]  [2700/6250]  eta: 0:16:27  lr: 0.000067  grad: 0.1192 (0.1326)  loss: 0.7974 (0.7998)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [50]  [2800/6250]  eta: 0:15:59  lr: 0.000067  grad: 0.1192 (0.1329)  loss: 0.7964 (0.7998)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [50]  [2900/6250]  eta: 0:15:30  lr: 0.000067  grad: 0.1240 (0.1331)  loss: 0.8036 (0.7998)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [50]  [3000/6250]  eta: 0:15:01  lr: 0.000067  grad: 0.1207 (0.1331)  loss: 0.8012 (0.7998)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [50]  [3100/6250]  eta: 0:14:33  lr: 0.000067  grad: 0.1174 (0.1331)  loss: 0.8011 (0.7998)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [50]  [3200/6250]  eta: 0:14:30  lr: 0.000067  grad: 0.1153 (0.1329)  loss: 0.7986 (0.7997)  time: 0.8742  data: 0.6048  max mem: 26157
Train: [50]  [3300/6250]  eta: 0:14:00  lr: 0.000067  grad: 0.1307 (0.1331)  loss: 0.7953 (0.7996)  time: 0.2743  data: 0.0002  max mem: 26157
Train: [50]  [3400/6250]  eta: 0:13:33  lr: 0.000067  grad: 0.1280 (0.1333)  loss: 0.7958 (0.7995)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [50]  [3500/6250]  eta: 0:13:03  lr: 0.000067  grad: 0.1383 (0.1337)  loss: 0.7891 (0.7994)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [50]  [3600/6250]  eta: 0:12:36  lr: 0.000066  grad: 0.1231 (0.1337)  loss: 0.7975 (0.7993)  time: 0.3139  data: 0.0378  max mem: 26157
Train: [50]  [3700/6250]  eta: 0:12:08  lr: 0.000066  grad: 0.1262 (0.1337)  loss: 0.7979 (0.7992)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [50]  [3800/6250]  eta: 0:11:38  lr: 0.000066  grad: 0.1306 (0.1338)  loss: 0.7971 (0.7990)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [50]  [3900/6250]  eta: 0:11:09  lr: 0.000066  grad: 0.1202 (0.1339)  loss: 0.7956 (0.7989)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [50]  [4000/6250]  eta: 0:10:39  lr: 0.000066  grad: 0.1314 (0.1339)  loss: 0.7928 (0.7987)  time: 0.2697  data: 0.0003  max mem: 26157
Train: [50]  [4100/6250]  eta: 0:10:10  lr: 0.000066  grad: 0.1231 (0.1341)  loss: 0.7886 (0.7986)  time: 0.2736  data: 0.0002  max mem: 26157
Train: [50]  [4200/6250]  eta: 0:09:41  lr: 0.000066  grad: 0.1315 (0.1341)  loss: 0.7948 (0.7985)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [50]  [4300/6250]  eta: 0:09:19  lr: 0.000066  grad: 0.1318 (0.1342)  loss: 0.7936 (0.7984)  time: 0.2743  data: 0.0002  max mem: 26157
Train: [50]  [4400/6250]  eta: 0:08:49  lr: 0.000066  grad: 0.1168 (0.1341)  loss: 0.8005 (0.7983)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [50]  [4500/6250]  eta: 0:08:20  lr: 0.000066  grad: 0.1245 (0.1340)  loss: 0.7972 (0.7983)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [50]  [4600/6250]  eta: 0:07:51  lr: 0.000066  grad: 0.1276 (0.1340)  loss: 0.8012 (0.7983)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [50]  [4700/6250]  eta: 0:07:22  lr: 0.000066  grad: 0.1200 (0.1339)  loss: 0.7928 (0.7984)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [50]  [4800/6250]  eta: 0:06:53  lr: 0.000066  grad: 0.1210 (0.1339)  loss: 0.8060 (0.7984)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [50]  [4900/6250]  eta: 0:06:24  lr: 0.000066  grad: 0.1182 (0.1339)  loss: 0.8003 (0.7984)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [50]  [5000/6250]  eta: 0:05:55  lr: 0.000066  grad: 0.1225 (0.1341)  loss: 0.7925 (0.7984)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [50]  [5100/6250]  eta: 0:05:26  lr: 0.000066  grad: 0.1220 (0.1341)  loss: 0.7996 (0.7984)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [50]  [5200/6250]  eta: 0:04:57  lr: 0.000066  grad: 0.1218 (0.1339)  loss: 0.7986 (0.7983)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [50]  [5300/6250]  eta: 0:04:29  lr: 0.000066  grad: 0.1239 (0.1338)  loss: 0.7945 (0.7984)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [50]  [5400/6250]  eta: 0:04:00  lr: 0.000066  grad: 0.1308 (0.1338)  loss: 0.7921 (0.7984)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [50]  [5500/6250]  eta: 0:03:32  lr: 0.000066  grad: 0.1221 (0.1340)  loss: 0.7971 (0.7984)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [50]  [5600/6250]  eta: 0:03:03  lr: 0.000066  grad: 0.1219 (0.1340)  loss: 0.8026 (0.7984)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [50]  [5700/6250]  eta: 0:02:35  lr: 0.000066  grad: 0.1248 (0.1341)  loss: 0.7976 (0.7984)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [50]  [5800/6250]  eta: 0:02:07  lr: 0.000066  grad: 0.1175 (0.1341)  loss: 0.8021 (0.7984)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [50]  [5900/6250]  eta: 0:01:38  lr: 0.000066  grad: 0.1287 (0.1342)  loss: 0.7936 (0.7984)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [50]  [6000/6250]  eta: 0:01:10  lr: 0.000066  grad: 0.1302 (0.1343)  loss: 0.7966 (0.7983)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [50]  [6100/6250]  eta: 0:00:42  lr: 0.000066  grad: 0.1222 (0.1342)  loss: 0.8003 (0.7984)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [50]  [6200/6250]  eta: 0:00:14  lr: 0.000066  grad: 0.1258 (0.1345)  loss: 0.7957 (0.7983)  time: 0.7128  data: 0.4313  max mem: 26157
Train: [50]  [6249/6250]  eta: 0:00:00  lr: 0.000066  grad: 0.1308 (0.1346)  loss: 0.7977 (0.7983)  time: 0.4723  data: 0.1976  max mem: 26157
Train: [50] Total time: 0:29:51 (0.2867 s / it)
Averaged stats: lr: 0.000066  grad: 0.1308 (0.1346)  loss: 0.7977 (0.7983)
Eval (hcp-train-subset): [50]  [ 0/62]  eta: 0:04:46  loss: 0.8307 (0.8307)  time: 4.6236  data: 4.5032  max mem: 26157
Eval (hcp-train-subset): [50]  [61/62]  eta: 0:00:00  loss: 0.8217 (0.8223)  time: 0.1465  data: 0.0634  max mem: 26157
Eval (hcp-train-subset): [50] Total time: 0:00:13 (0.2202 s / it)
Averaged stats (hcp-train-subset): loss: 0.8217 (0.8223)
Making plots (hcp-train-subset): example=36
Eval (hcp-val): [50]  [ 0/62]  eta: 0:04:52  loss: 0.8234 (0.8234)  time: 4.7252  data: 4.6416  max mem: 26157
Eval (hcp-val): [50]  [61/62]  eta: 0:00:00  loss: 0.8235 (0.8251)  time: 0.1146  data: 0.0318  max mem: 26157
Eval (hcp-val): [50] Total time: 0:00:13 (0.2104 s / it)
Averaged stats (hcp-val): loss: 0.8235 (0.8251)
Making plots (hcp-val): example=39
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [51]  [   0/6250]  eta: 8:20:08  lr: 0.000066  grad: 0.2176 (0.2176)  loss: 0.7914 (0.7914)  time: 4.8013  data: 4.4505  max mem: 26157
Train: [51]  [ 100/6250]  eta: 0:33:56  lr: 0.000066  grad: 0.1245 (0.1457)  loss: 0.8146 (0.8204)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [51]  [ 200/6250]  eta: 0:30:19  lr: 0.000066  grad: 0.1370 (0.1411)  loss: 0.8082 (0.8126)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [51]  [ 300/6250]  eta: 0:28:49  lr: 0.000065  grad: 0.1295 (0.1380)  loss: 0.8120 (0.8093)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [51]  [ 400/6250]  eta: 0:27:49  lr: 0.000065  grad: 0.1258 (0.1366)  loss: 0.7906 (0.8070)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [51]  [ 500/6250]  eta: 0:27:00  lr: 0.000065  grad: 0.1155 (0.1346)  loss: 0.8047 (0.8063)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [51]  [ 600/6250]  eta: 0:26:23  lr: 0.000065  grad: 0.1284 (0.1339)  loss: 0.8038 (0.8053)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [51]  [ 700/6250]  eta: 0:25:46  lr: 0.000065  grad: 0.1218 (0.1344)  loss: 0.8039 (0.8043)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [51]  [ 800/6250]  eta: 0:25:12  lr: 0.000065  grad: 0.1202 (0.1343)  loss: 0.7986 (0.8038)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [51]  [ 900/6250]  eta: 0:24:39  lr: 0.000065  grad: 0.1206 (0.1337)  loss: 0.8026 (0.8035)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [51]  [1000/6250]  eta: 0:24:09  lr: 0.000065  grad: 0.1090 (0.1325)  loss: 0.8076 (0.8033)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [51]  [1100/6250]  eta: 0:23:38  lr: 0.000065  grad: 0.1247 (0.1318)  loss: 0.7934 (0.8025)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [51]  [1200/6250]  eta: 0:23:08  lr: 0.000065  grad: 0.1198 (0.1316)  loss: 0.7982 (0.8020)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [51]  [1300/6250]  eta: 0:22:38  lr: 0.000065  grad: 0.1197 (0.1312)  loss: 0.7996 (0.8017)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [51]  [1400/6250]  eta: 0:22:09  lr: 0.000065  grad: 0.1210 (0.1311)  loss: 0.7982 (0.8015)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [51]  [1500/6250]  eta: 0:21:41  lr: 0.000065  grad: 0.1195 (0.1311)  loss: 0.7961 (0.8011)  time: 0.2764  data: 0.0003  max mem: 26157
Train: [51]  [1600/6250]  eta: 0:21:27  lr: 0.000065  grad: 0.1148 (0.1313)  loss: 0.8029 (0.8009)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [51]  [1700/6250]  eta: 0:21:10  lr: 0.000065  grad: 0.1174 (0.1316)  loss: 0.7976 (0.8007)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [51]  [1800/6250]  eta: 0:20:45  lr: 0.000065  grad: 0.1232 (0.1320)  loss: 0.7965 (0.8004)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [51]  [1900/6250]  eta: 0:20:27  lr: 0.000065  grad: 0.1410 (0.1323)  loss: 0.7963 (0.8002)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [51]  [2000/6250]  eta: 0:20:06  lr: 0.000065  grad: 0.1300 (0.1322)  loss: 0.7980 (0.8001)  time: 0.3112  data: 0.0396  max mem: 26157
Train: [51]  [2100/6250]  eta: 0:19:36  lr: 0.000065  grad: 0.1271 (0.1323)  loss: 0.7932 (0.8000)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [51]  [2200/6250]  eta: 0:19:14  lr: 0.000065  grad: 0.1171 (0.1324)  loss: 0.8023 (0.7999)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [51]  [2300/6250]  eta: 0:18:43  lr: 0.000065  grad: 0.1142 (0.1323)  loss: 0.7988 (0.7998)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [51]  [2400/6250]  eta: 0:18:13  lr: 0.000065  grad: 0.1281 (0.1324)  loss: 0.7967 (0.7998)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [51]  [2500/6250]  eta: 0:17:42  lr: 0.000065  grad: 0.1204 (0.1322)  loss: 0.8003 (0.7997)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [51]  [2600/6250]  eta: 0:17:12  lr: 0.000065  grad: 0.1253 (0.1324)  loss: 0.8019 (0.7997)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [51]  [2700/6250]  eta: 0:16:42  lr: 0.000065  grad: 0.1235 (0.1322)  loss: 0.7969 (0.7997)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [51]  [2800/6250]  eta: 0:16:13  lr: 0.000065  grad: 0.1184 (0.1322)  loss: 0.7989 (0.7996)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [51]  [2900/6250]  eta: 0:15:45  lr: 0.000065  grad: 0.1196 (0.1321)  loss: 0.8044 (0.7996)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [51]  [3000/6250]  eta: 0:15:16  lr: 0.000065  grad: 0.1165 (0.1321)  loss: 0.7937 (0.7994)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [51]  [3100/6250]  eta: 0:14:46  lr: 0.000065  grad: 0.1256 (0.1323)  loss: 0.7954 (0.7994)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [51]  [3200/6250]  eta: 0:14:17  lr: 0.000065  grad: 0.1155 (0.1323)  loss: 0.7980 (0.7993)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [51]  [3300/6250]  eta: 0:13:48  lr: 0.000065  grad: 0.1205 (0.1322)  loss: 0.7944 (0.7992)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [51]  [3400/6250]  eta: 0:13:19  lr: 0.000064  grad: 0.1225 (0.1320)  loss: 0.7961 (0.7991)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [51]  [3500/6250]  eta: 0:12:50  lr: 0.000064  grad: 0.1167 (0.1321)  loss: 0.7978 (0.7990)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [51]  [3600/6250]  eta: 0:12:21  lr: 0.000064  grad: 0.1194 (0.1322)  loss: 0.8040 (0.7990)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [51]  [3700/6250]  eta: 0:11:53  lr: 0.000064  grad: 0.1285 (0.1322)  loss: 0.7948 (0.7989)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [51]  [3800/6250]  eta: 0:11:28  lr: 0.000064  grad: 0.1262 (0.1321)  loss: 0.7950 (0.7989)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [51]  [3900/6250]  eta: 0:11:00  lr: 0.000064  grad: 0.1208 (0.1323)  loss: 0.8012 (0.7988)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [51]  [4000/6250]  eta: 0:10:31  lr: 0.000064  grad: 0.1348 (0.1325)  loss: 0.7946 (0.7988)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [51]  [4100/6250]  eta: 0:10:04  lr: 0.000064  grad: 0.1309 (0.1325)  loss: 0.7992 (0.7988)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [51]  [4200/6250]  eta: 0:09:40  lr: 0.000064  grad: 0.1229 (0.1325)  loss: 0.8002 (0.7988)  time: 0.5639  data: 0.2899  max mem: 26157
Train: [51]  [4300/6250]  eta: 0:09:13  lr: 0.000064  grad: 0.1285 (0.1326)  loss: 0.7953 (0.7988)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [51]  [4400/6250]  eta: 0:08:50  lr: 0.000064  grad: 0.1260 (0.1327)  loss: 0.7931 (0.7987)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [51]  [4500/6250]  eta: 0:08:21  lr: 0.000064  grad: 0.1214 (0.1328)  loss: 0.7975 (0.7987)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [51]  [4600/6250]  eta: 0:07:52  lr: 0.000064  grad: 0.1230 (0.1328)  loss: 0.7881 (0.7986)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [51]  [4700/6250]  eta: 0:07:22  lr: 0.000064  grad: 0.1306 (0.1330)  loss: 0.7938 (0.7985)  time: 0.2693  data: 0.0001  max mem: 26157
Train: [51]  [4800/6250]  eta: 0:06:53  lr: 0.000064  grad: 0.1328 (0.1332)  loss: 0.7890 (0.7984)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [51]  [4900/6250]  eta: 0:06:24  lr: 0.000064  grad: 0.1295 (0.1336)  loss: 0.7984 (0.7983)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [51]  [5000/6250]  eta: 0:05:55  lr: 0.000064  grad: 0.1343 (0.1338)  loss: 0.7868 (0.7982)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [51]  [5100/6250]  eta: 0:05:27  lr: 0.000064  grad: 0.1254 (0.1339)  loss: 0.7916 (0.7981)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [51]  [5200/6250]  eta: 0:04:58  lr: 0.000064  grad: 0.1330 (0.1341)  loss: 0.7843 (0.7979)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [51]  [5300/6250]  eta: 0:04:29  lr: 0.000064  grad: 0.1348 (0.1343)  loss: 0.7898 (0.7978)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [51]  [5400/6250]  eta: 0:04:01  lr: 0.000064  grad: 0.1377 (0.1345)  loss: 0.7966 (0.7977)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [51]  [5500/6250]  eta: 0:03:32  lr: 0.000064  grad: 0.1295 (0.1347)  loss: 0.7955 (0.7975)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [51]  [5600/6250]  eta: 0:03:04  lr: 0.000064  grad: 0.1306 (0.1349)  loss: 0.7888 (0.7974)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [51]  [5700/6250]  eta: 0:02:36  lr: 0.000064  grad: 0.1355 (0.1351)  loss: 0.7893 (0.7973)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [51]  [5800/6250]  eta: 0:02:07  lr: 0.000064  grad: 0.1262 (0.1351)  loss: 0.7907 (0.7972)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [51]  [5900/6250]  eta: 0:01:39  lr: 0.000064  grad: 0.1347 (0.1351)  loss: 0.7903 (0.7972)  time: 0.2751  data: 0.0002  max mem: 26157
Train: [51]  [6000/6250]  eta: 0:01:10  lr: 0.000064  grad: 0.1192 (0.1352)  loss: 0.7897 (0.7971)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [51]  [6100/6250]  eta: 0:00:42  lr: 0.000064  grad: 0.1213 (0.1352)  loss: 0.7989 (0.7971)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [51]  [6200/6250]  eta: 0:00:14  lr: 0.000064  grad: 0.1238 (0.1351)  loss: 0.7879 (0.7971)  time: 0.2769  data: 0.0002  max mem: 26157
Train: [51]  [6249/6250]  eta: 0:00:00  lr: 0.000064  grad: 0.1281 (0.1351)  loss: 0.7912 (0.7970)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [51] Total time: 0:29:35 (0.2841 s / it)
Averaged stats: lr: 0.000064  grad: 0.1281 (0.1351)  loss: 0.7912 (0.7970)
Eval (hcp-train-subset): [51]  [ 0/62]  eta: 0:03:38  loss: 0.8349 (0.8349)  time: 3.5191  data: 3.4074  max mem: 26157
Eval (hcp-train-subset): [51]  [61/62]  eta: 0:00:00  loss: 0.8209 (0.8229)  time: 0.1363  data: 0.0517  max mem: 26157
Eval (hcp-train-subset): [51] Total time: 0:00:12 (0.2081 s / it)
Averaged stats (hcp-train-subset): loss: 0.8209 (0.8229)
Making plots (hcp-train-subset): example=59
Eval (hcp-val): [51]  [ 0/62]  eta: 0:04:39  loss: 0.8218 (0.8218)  time: 4.5029  data: 4.3941  max mem: 26157
Eval (hcp-val): [51]  [61/62]  eta: 0:00:00  loss: 0.8255 (0.8263)  time: 0.1366  data: 0.0537  max mem: 26157
Eval (hcp-val): [51] Total time: 0:00:13 (0.2141 s / it)
Averaged stats (hcp-val): loss: 0.8255 (0.8263)
Making plots (hcp-val): example=9
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [52]  [   0/6250]  eta: 8:35:51  lr: 0.000064  grad: 0.1210 (0.1210)  loss: 0.8227 (0.8227)  time: 4.9522  data: 4.6101  max mem: 26157
Train: [52]  [ 100/6250]  eta: 0:33:04  lr: 0.000063  grad: 0.1371 (0.1805)  loss: 0.7902 (0.8050)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [52]  [ 200/6250]  eta: 0:29:47  lr: 0.000063  grad: 0.1230 (0.1607)  loss: 0.7981 (0.8024)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [52]  [ 300/6250]  eta: 0:28:26  lr: 0.000063  grad: 0.1341 (0.1522)  loss: 0.7887 (0.8014)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [52]  [ 400/6250]  eta: 0:27:34  lr: 0.000063  grad: 0.1388 (0.1504)  loss: 0.7949 (0.8013)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [52]  [ 500/6250]  eta: 0:26:53  lr: 0.000063  grad: 0.1390 (0.1484)  loss: 0.8024 (0.8008)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [52]  [ 600/6250]  eta: 0:26:15  lr: 0.000063  grad: 0.1357 (0.1465)  loss: 0.8059 (0.8013)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [52]  [ 700/6250]  eta: 0:25:45  lr: 0.000063  grad: 0.1332 (0.1453)  loss: 0.7981 (0.8010)  time: 0.2810  data: 0.0002  max mem: 26157
Train: [52]  [ 800/6250]  eta: 0:28:39  lr: 0.000063  grad: 0.1315 (0.1439)  loss: 0.7962 (0.8004)  time: 0.4388  data: 0.1648  max mem: 26157
Train: [52]  [ 900/6250]  eta: 0:27:42  lr: 0.000063  grad: 0.1278 (0.1432)  loss: 0.8007 (0.8002)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [52]  [1000/6250]  eta: 0:26:50  lr: 0.000063  grad: 0.1311 (0.1434)  loss: 0.8010 (0.8000)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [52]  [1100/6250]  eta: 0:26:47  lr: 0.000063  grad: 0.1280 (0.1431)  loss: 0.8014 (0.7998)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [52]  [1200/6250]  eta: 0:25:59  lr: 0.000063  grad: 0.1338 (0.1426)  loss: 0.7957 (0.7998)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [52]  [1300/6250]  eta: 0:25:14  lr: 0.000063  grad: 0.1353 (0.1429)  loss: 0.7938 (0.7995)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [52]  [1400/6250]  eta: 0:24:31  lr: 0.000063  grad: 0.1201 (0.1423)  loss: 0.7973 (0.7991)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [52]  [1500/6250]  eta: 0:23:51  lr: 0.000063  grad: 0.1315 (0.1427)  loss: 0.7995 (0.7988)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [52]  [1600/6250]  eta: 0:23:39  lr: 0.000063  grad: 0.1336 (0.1432)  loss: 0.7941 (0.7985)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [52]  [1700/6250]  eta: 0:22:59  lr: 0.000063  grad: 0.1419 (0.1438)  loss: 0.7980 (0.7984)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [52]  [1800/6250]  eta: 0:22:49  lr: 0.000063  grad: 0.1278 (0.1442)  loss: 0.7975 (0.7982)  time: 0.8397  data: 0.5642  max mem: 26157
Train: [52]  [1900/6250]  eta: 0:22:10  lr: 0.000063  grad: 0.1346 (0.1439)  loss: 0.7927 (0.7982)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [52]  [2000/6250]  eta: 0:21:32  lr: 0.000063  grad: 0.1305 (0.1441)  loss: 0.7940 (0.7981)  time: 0.2753  data: 0.0002  max mem: 26157
Train: [52]  [2100/6250]  eta: 0:20:55  lr: 0.000063  grad: 0.1525 (0.1450)  loss: 0.7911 (0.7978)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [52]  [2200/6250]  eta: 0:20:19  lr: 0.000063  grad: 0.1294 (0.1450)  loss: 0.7953 (0.7976)  time: 0.2739  data: 0.0003  max mem: 26157
Train: [52]  [2300/6250]  eta: 0:20:02  lr: 0.000063  grad: 0.1298 (0.1448)  loss: 0.7986 (0.7975)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [52]  [2400/6250]  eta: 0:19:26  lr: 0.000063  grad: 0.1311 (0.1454)  loss: 0.7970 (0.7973)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [52]  [2500/6250]  eta: 0:18:51  lr: 0.000063  grad: 0.1539 (0.1456)  loss: 0.7828 (0.7972)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [52]  [2600/6250]  eta: 0:18:16  lr: 0.000063  grad: 0.1356 (0.1456)  loss: 0.7879 (0.7969)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [52]  [2700/6250]  eta: 0:17:42  lr: 0.000063  grad: 0.1288 (0.1454)  loss: 0.7919 (0.7968)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [52]  [2800/6250]  eta: 0:17:08  lr: 0.000063  grad: 0.1293 (0.1455)  loss: 0.7922 (0.7966)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [52]  [2900/6250]  eta: 0:16:35  lr: 0.000063  grad: 0.1383 (0.1458)  loss: 0.7893 (0.7964)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [52]  [3000/6250]  eta: 0:16:03  lr: 0.000063  grad: 0.1258 (0.1460)  loss: 0.7934 (0.7962)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [52]  [3100/6250]  eta: 0:15:31  lr: 0.000063  grad: 0.1329 (0.1460)  loss: 0.8041 (0.7962)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [52]  [3200/6250]  eta: 0:14:58  lr: 0.000062  grad: 0.1362 (0.1461)  loss: 0.7888 (0.7961)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [52]  [3300/6250]  eta: 0:14:28  lr: 0.000062  grad: 0.1321 (0.1459)  loss: 0.7871 (0.7960)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [52]  [3400/6250]  eta: 0:13:56  lr: 0.000062  grad: 0.1255 (0.1458)  loss: 0.7927 (0.7959)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [52]  [3500/6250]  eta: 0:13:25  lr: 0.000062  grad: 0.1427 (0.1456)  loss: 0.7926 (0.7958)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [52]  [3600/6250]  eta: 0:12:54  lr: 0.000062  grad: 0.1228 (0.1453)  loss: 0.7938 (0.7957)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [52]  [3700/6250]  eta: 0:12:24  lr: 0.000062  grad: 0.1238 (0.1450)  loss: 0.7932 (0.7956)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [52]  [3800/6250]  eta: 0:11:53  lr: 0.000062  grad: 0.1201 (0.1447)  loss: 0.7995 (0.7956)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [52]  [3900/6250]  eta: 0:11:23  lr: 0.000062  grad: 0.1438 (0.1445)  loss: 0.7972 (0.7956)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [52]  [4000/6250]  eta: 0:10:52  lr: 0.000062  grad: 0.1258 (0.1442)  loss: 0.7915 (0.7956)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [52]  [4100/6250]  eta: 0:10:22  lr: 0.000062  grad: 0.1256 (0.1440)  loss: 0.7981 (0.7956)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [52]  [4200/6250]  eta: 0:09:52  lr: 0.000062  grad: 0.1297 (0.1438)  loss: 0.7934 (0.7955)  time: 0.2724  data: 0.0003  max mem: 26157
Train: [52]  [4300/6250]  eta: 0:09:23  lr: 0.000062  grad: 0.1276 (0.1435)  loss: 0.7915 (0.7954)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [52]  [4400/6250]  eta: 0:08:53  lr: 0.000062  grad: 0.1296 (0.1433)  loss: 0.7975 (0.7954)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [52]  [4500/6250]  eta: 0:08:23  lr: 0.000062  grad: 0.1299 (0.1430)  loss: 0.7925 (0.7954)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [52]  [4600/6250]  eta: 0:07:54  lr: 0.000062  grad: 0.1246 (0.1429)  loss: 0.7896 (0.7954)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [52]  [4700/6250]  eta: 0:07:25  lr: 0.000062  grad: 0.1325 (0.1430)  loss: 0.7925 (0.7955)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [52]  [4800/6250]  eta: 0:06:55  lr: 0.000062  grad: 0.1266 (0.1432)  loss: 0.7989 (0.7956)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [52]  [4900/6250]  eta: 0:06:26  lr: 0.000062  grad: 0.1373 (0.1433)  loss: 0.7939 (0.7956)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [52]  [5000/6250]  eta: 0:05:57  lr: 0.000062  grad: 0.1350 (0.1435)  loss: 0.7998 (0.7957)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [52]  [5100/6250]  eta: 0:05:28  lr: 0.000062  grad: 0.1234 (0.1434)  loss: 0.7999 (0.7957)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [52]  [5200/6250]  eta: 0:05:00  lr: 0.000062  grad: 0.1328 (0.1434)  loss: 0.7983 (0.7957)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [52]  [5300/6250]  eta: 0:04:33  lr: 0.000062  grad: 0.1340 (0.1433)  loss: 0.7908 (0.7957)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [52]  [5400/6250]  eta: 0:04:04  lr: 0.000062  grad: 0.1201 (0.1432)  loss: 0.8034 (0.7958)  time: 0.2716  data: 0.0001  max mem: 26157
Train: [52]  [5500/6250]  eta: 0:03:35  lr: 0.000062  grad: 0.1303 (0.1431)  loss: 0.7978 (0.7958)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [52]  [5600/6250]  eta: 0:03:07  lr: 0.000062  grad: 0.1293 (0.1430)  loss: 0.7964 (0.7958)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [52]  [5700/6250]  eta: 0:02:38  lr: 0.000062  grad: 0.1200 (0.1429)  loss: 0.8036 (0.7958)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [52]  [5800/6250]  eta: 0:02:09  lr: 0.000062  grad: 0.1299 (0.1428)  loss: 0.7980 (0.7959)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [52]  [5900/6250]  eta: 0:01:40  lr: 0.000062  grad: 0.1320 (0.1428)  loss: 0.7976 (0.7959)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [52]  [6000/6250]  eta: 0:01:11  lr: 0.000062  grad: 0.1291 (0.1426)  loss: 0.8009 (0.7960)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [52]  [6100/6250]  eta: 0:00:43  lr: 0.000062  grad: 0.1296 (0.1424)  loss: 0.8008 (0.7961)  time: 0.2757  data: 0.0002  max mem: 26157
Train: [52]  [6200/6250]  eta: 0:00:14  lr: 0.000061  grad: 0.1301 (0.1424)  loss: 0.8003 (0.7961)  time: 0.4998  data: 0.2279  max mem: 26157
Train: [52]  [6249/6250]  eta: 0:00:00  lr: 0.000061  grad: 0.1298 (0.1425)  loss: 0.8001 (0.7961)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [52] Total time: 0:30:06 (0.2891 s / it)
Averaged stats: lr: 0.000061  grad: 0.1298 (0.1425)  loss: 0.8001 (0.7961)
Eval (hcp-train-subset): [52]  [ 0/62]  eta: 0:05:08  loss: 0.8337 (0.8337)  time: 4.9682  data: 4.8847  max mem: 26157
Eval (hcp-train-subset): [52]  [61/62]  eta: 0:00:00  loss: 0.8184 (0.8204)  time: 0.1376  data: 0.0546  max mem: 26157
Eval (hcp-train-subset): [52] Total time: 0:00:13 (0.2163 s / it)
Averaged stats (hcp-train-subset): loss: 0.8184 (0.8204)
Making plots (hcp-train-subset): example=38
Eval (hcp-val): [52]  [ 0/62]  eta: 0:03:16  loss: 0.8197 (0.8197)  time: 3.1741  data: 3.0883  max mem: 26157
Eval (hcp-val): [52]  [61/62]  eta: 0:00:00  loss: 0.8242 (0.8258)  time: 0.1333  data: 0.0507  max mem: 26157
Eval (hcp-val): [52] Total time: 0:00:13 (0.2103 s / it)
Averaged stats (hcp-val): loss: 0.8242 (0.8258)
Making plots (hcp-val): example=39
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [53]  [   0/6250]  eta: 11:49:09  lr: 0.000061  grad: 0.1262 (0.1262)  loss: 0.8527 (0.8527)  time: 6.8080  data: 6.5301  max mem: 26157
Train: [53]  [ 100/6250]  eta: 0:35:08  lr: 0.000061  grad: 0.1294 (0.1828)  loss: 0.8045 (0.8172)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [53]  [ 200/6250]  eta: 0:30:58  lr: 0.000061  grad: 0.1303 (0.1684)  loss: 0.8061 (0.8089)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [53]  [ 300/6250]  eta: 0:29:16  lr: 0.000061  grad: 0.1247 (0.1574)  loss: 0.7971 (0.8069)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [53]  [ 400/6250]  eta: 0:28:10  lr: 0.000061  grad: 0.1219 (0.1542)  loss: 0.8043 (0.8060)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [53]  [ 500/6250]  eta: 0:27:19  lr: 0.000061  grad: 0.1311 (0.1505)  loss: 0.8041 (0.8054)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [53]  [ 600/6250]  eta: 0:26:39  lr: 0.000061  grad: 0.1370 (0.1504)  loss: 0.7966 (0.8044)  time: 0.2777  data: 0.0002  max mem: 26157
Train: [53]  [ 700/6250]  eta: 0:27:16  lr: 0.000061  grad: 0.1387 (0.1503)  loss: 0.7931 (0.8035)  time: 0.6934  data: 0.4149  max mem: 26157
Train: [53]  [ 800/6250]  eta: 0:26:31  lr: 0.000061  grad: 0.1337 (0.1514)  loss: 0.7862 (0.8025)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [53]  [ 900/6250]  eta: 0:26:10  lr: 0.000061  grad: 0.1211 (0.1503)  loss: 0.7952 (0.8022)  time: 0.4300  data: 0.1563  max mem: 26157
Train: [53]  [1000/6250]  eta: 0:25:29  lr: 0.000061  grad: 0.1370 (0.1502)  loss: 0.7964 (0.8013)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [53]  [1100/6250]  eta: 0:25:05  lr: 0.000061  grad: 0.1393 (0.1489)  loss: 0.7824 (0.8003)  time: 0.4261  data: 0.1467  max mem: 26157
Train: [53]  [1200/6250]  eta: 0:25:53  lr: 0.000061  grad: 0.1296 (0.1483)  loss: 0.7930 (0.7996)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [53]  [1300/6250]  eta: 0:25:08  lr: 0.000061  grad: 0.1355 (0.1485)  loss: 0.7881 (0.7990)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [53]  [1400/6250]  eta: 0:24:25  lr: 0.000061  grad: 0.1328 (0.1484)  loss: 0.7945 (0.7984)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [53]  [1500/6250]  eta: 0:23:45  lr: 0.000061  grad: 0.1373 (0.1500)  loss: 0.7847 (0.7978)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [53]  [1600/6250]  eta: 0:23:07  lr: 0.000061  grad: 0.1290 (0.1500)  loss: 0.7895 (0.7973)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [53]  [1700/6250]  eta: 0:22:30  lr: 0.000061  grad: 0.1268 (0.1499)  loss: 0.7887 (0.7969)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [53]  [1800/6250]  eta: 0:21:53  lr: 0.000061  grad: 0.1311 (0.1502)  loss: 0.7924 (0.7965)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [53]  [1900/6250]  eta: 0:21:18  lr: 0.000061  grad: 0.1414 (0.1507)  loss: 0.7923 (0.7962)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [53]  [2000/6250]  eta: 0:20:44  lr: 0.000061  grad: 0.1319 (0.1509)  loss: 0.7888 (0.7960)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [53]  [2100/6250]  eta: 0:20:10  lr: 0.000061  grad: 0.1371 (0.1510)  loss: 0.7881 (0.7957)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [53]  [2200/6250]  eta: 0:19:37  lr: 0.000061  grad: 0.1246 (0.1508)  loss: 0.7931 (0.7955)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [53]  [2300/6250]  eta: 0:19:04  lr: 0.000061  grad: 0.1352 (0.1506)  loss: 0.7896 (0.7954)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [53]  [2400/6250]  eta: 0:18:33  lr: 0.000061  grad: 0.1274 (0.1508)  loss: 0.7928 (0.7952)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [53]  [2500/6250]  eta: 0:18:01  lr: 0.000061  grad: 0.1285 (0.1509)  loss: 0.7919 (0.7949)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [53]  [2600/6250]  eta: 0:17:41  lr: 0.000061  grad: 0.1252 (0.1508)  loss: 0.7926 (0.7947)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [53]  [2700/6250]  eta: 0:17:09  lr: 0.000061  grad: 0.1321 (0.1506)  loss: 0.7969 (0.7947)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [53]  [2800/6250]  eta: 0:16:39  lr: 0.000061  grad: 0.1283 (0.1506)  loss: 0.7974 (0.7946)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [53]  [2900/6250]  eta: 0:16:08  lr: 0.000061  grad: 0.1389 (0.1508)  loss: 0.7898 (0.7945)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [53]  [3000/6250]  eta: 0:15:38  lr: 0.000060  grad: 0.1222 (0.1505)  loss: 0.7958 (0.7944)  time: 0.2771  data: 0.0002  max mem: 26157
Train: [53]  [3100/6250]  eta: 0:15:15  lr: 0.000060  grad: 0.1278 (0.1507)  loss: 0.7905 (0.7943)  time: 0.2759  data: 0.0004  max mem: 26157
Train: [53]  [3200/6250]  eta: 0:14:44  lr: 0.000060  grad: 0.1258 (0.1504)  loss: 0.7983 (0.7943)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [53]  [3300/6250]  eta: 0:14:21  lr: 0.000060  grad: 0.1317 (0.1505)  loss: 0.7956 (0.7943)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [53]  [3400/6250]  eta: 0:13:50  lr: 0.000060  grad: 0.1301 (0.1501)  loss: 0.7985 (0.7944)  time: 0.2769  data: 0.0002  max mem: 26157
Train: [53]  [3500/6250]  eta: 0:13:19  lr: 0.000060  grad: 0.1409 (0.1503)  loss: 0.7930 (0.7945)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [53]  [3600/6250]  eta: 0:13:04  lr: 0.000060  grad: 0.1372 (0.1502)  loss: 0.7943 (0.7945)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [53]  [3700/6250]  eta: 0:12:33  lr: 0.000060  grad: 0.1172 (0.1501)  loss: 0.7946 (0.7945)  time: 0.2792  data: 0.0002  max mem: 26157
Train: [53]  [3800/6250]  eta: 0:12:07  lr: 0.000060  grad: 0.1436 (0.1501)  loss: 0.7897 (0.7945)  time: 0.5775  data: 0.3019  max mem: 26157
Train: [53]  [3900/6250]  eta: 0:11:36  lr: 0.000060  grad: 0.1369 (0.1502)  loss: 0.7929 (0.7945)  time: 0.2739  data: 0.0002  max mem: 26157
Train: [53]  [4000/6250]  eta: 0:11:05  lr: 0.000060  grad: 0.1255 (0.1503)  loss: 0.7942 (0.7944)  time: 0.2693  data: 0.0003  max mem: 26157
Train: [53]  [4100/6250]  eta: 0:10:34  lr: 0.000060  grad: 0.1330 (0.1501)  loss: 0.7895 (0.7944)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [53]  [4200/6250]  eta: 0:10:03  lr: 0.000060  grad: 0.1275 (0.1502)  loss: 0.7987 (0.7944)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [53]  [4300/6250]  eta: 0:09:33  lr: 0.000060  grad: 0.1311 (0.1499)  loss: 0.7934 (0.7944)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [53]  [4400/6250]  eta: 0:09:02  lr: 0.000060  grad: 0.1500 (0.1500)  loss: 0.7970 (0.7944)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [53]  [4500/6250]  eta: 0:08:32  lr: 0.000060  grad: 0.1305 (0.1499)  loss: 0.7900 (0.7944)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [53]  [4600/6250]  eta: 0:08:02  lr: 0.000060  grad: 0.1295 (0.1498)  loss: 0.7891 (0.7944)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [53]  [4700/6250]  eta: 0:07:32  lr: 0.000060  grad: 0.1366 (0.1498)  loss: 0.7919 (0.7943)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [53]  [4800/6250]  eta: 0:07:02  lr: 0.000060  grad: 0.1398 (0.1498)  loss: 0.7936 (0.7943)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [53]  [4900/6250]  eta: 0:06:32  lr: 0.000060  grad: 0.1389 (0.1499)  loss: 0.7840 (0.7942)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [53]  [5000/6250]  eta: 0:06:03  lr: 0.000060  grad: 0.1424 (0.1505)  loss: 0.7833 (0.7940)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [53]  [5100/6250]  eta: 0:05:33  lr: 0.000060  grad: 0.1348 (0.1505)  loss: 0.7908 (0.7939)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [53]  [5200/6250]  eta: 0:05:04  lr: 0.000060  grad: 0.1313 (0.1504)  loss: 0.7886 (0.7938)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [53]  [5300/6250]  eta: 0:04:34  lr: 0.000060  grad: 0.1443 (0.1505)  loss: 0.7820 (0.7937)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [53]  [5400/6250]  eta: 0:04:05  lr: 0.000060  grad: 0.1254 (0.1504)  loss: 0.7866 (0.7937)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [53]  [5500/6250]  eta: 0:03:36  lr: 0.000060  grad: 0.1247 (0.1503)  loss: 0.7902 (0.7937)  time: 0.4412  data: 0.1704  max mem: 26157
Train: [53]  [5600/6250]  eta: 0:03:07  lr: 0.000060  grad: 0.1398 (0.1504)  loss: 0.7963 (0.7937)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [53]  [5700/6250]  eta: 0:02:38  lr: 0.000060  grad: 0.1349 (0.1505)  loss: 0.7949 (0.7936)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [53]  [5800/6250]  eta: 0:02:09  lr: 0.000060  grad: 0.1348 (0.1505)  loss: 0.7887 (0.7936)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [53]  [5900/6250]  eta: 0:01:40  lr: 0.000060  grad: 0.1322 (0.1504)  loss: 0.7933 (0.7935)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [53]  [6000/6250]  eta: 0:01:11  lr: 0.000059  grad: 0.1343 (0.1505)  loss: 0.7907 (0.7935)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [53]  [6100/6250]  eta: 0:00:43  lr: 0.000059  grad: 0.1311 (0.1506)  loss: 0.7884 (0.7935)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [53]  [6200/6250]  eta: 0:00:14  lr: 0.000059  grad: 0.1322 (0.1505)  loss: 0.7961 (0.7934)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [53]  [6249/6250]  eta: 0:00:00  lr: 0.000059  grad: 0.1318 (0.1505)  loss: 0.7901 (0.7934)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [53] Total time: 0:29:59 (0.2880 s / it)
Averaged stats: lr: 0.000059  grad: 0.1318 (0.1505)  loss: 0.7901 (0.7934)
Eval (hcp-train-subset): [53]  [ 0/62]  eta: 0:06:02  loss: 0.8361 (0.8361)  time: 5.8547  data: 5.7717  max mem: 26157
Eval (hcp-train-subset): [53]  [61/62]  eta: 0:00:00  loss: 0.8195 (0.8211)  time: 0.1156  data: 0.0307  max mem: 26157
Eval (hcp-train-subset): [53] Total time: 0:00:13 (0.2115 s / it)
Averaged stats (hcp-train-subset): loss: 0.8195 (0.8211)
Making plots (hcp-train-subset): example=43
Eval (hcp-val): [53]  [ 0/62]  eta: 0:04:42  loss: 0.8225 (0.8225)  time: 4.5593  data: 4.4638  max mem: 26157
Eval (hcp-val): [53]  [61/62]  eta: 0:00:00  loss: 0.8244 (0.8258)  time: 0.1169  data: 0.0337  max mem: 26157
Eval (hcp-val): [53] Total time: 0:00:13 (0.2149 s / it)
Averaged stats (hcp-val): loss: 0.8244 (0.8258)
Making plots (hcp-val): example=23
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [54]  [   0/6250]  eta: 9:51:52  lr: 0.000059  grad: 0.1924 (0.1924)  loss: 0.8357 (0.8357)  time: 5.6820  data: 5.4022  max mem: 26157
Train: [54]  [ 100/6250]  eta: 0:33:52  lr: 0.000059  grad: 0.1393 (0.1632)  loss: 0.7875 (0.8005)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [54]  [ 200/6250]  eta: 0:30:23  lr: 0.000059  grad: 0.1247 (0.1525)  loss: 0.8051 (0.7980)  time: 0.2708  data: 0.0003  max mem: 26157
Train: [54]  [ 300/6250]  eta: 0:31:31  lr: 0.000059  grad: 0.1261 (0.1482)  loss: 0.7985 (0.7984)  time: 0.6551  data: 0.3400  max mem: 26157
Train: [54]  [ 400/6250]  eta: 0:29:58  lr: 0.000059  grad: 0.1352 (0.1473)  loss: 0.8035 (0.7974)  time: 0.2773  data: 0.0002  max mem: 26157
Train: [54]  [ 500/6250]  eta: 0:29:12  lr: 0.000059  grad: 0.1218 (0.1462)  loss: 0.8012 (0.7975)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [54]  [ 600/6250]  eta: 0:28:21  lr: 0.000059  grad: 0.1395 (0.1474)  loss: 0.8015 (0.7976)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [54]  [ 700/6250]  eta: 0:27:30  lr: 0.000059  grad: 0.1297 (0.1477)  loss: 0.8025 (0.7973)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [54]  [ 800/6250]  eta: 0:27:43  lr: 0.000059  grad: 0.1278 (0.1471)  loss: 0.7967 (0.7976)  time: 0.4012  data: 0.1213  max mem: 26157
Train: [54]  [ 900/6250]  eta: 0:26:56  lr: 0.000059  grad: 0.1172 (0.1458)  loss: 0.8045 (0.7975)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [54]  [1000/6250]  eta: 0:26:46  lr: 0.000059  grad: 0.1266 (0.1445)  loss: 0.8006 (0.7975)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [54]  [1100/6250]  eta: 0:27:11  lr: 0.000059  grad: 0.1180 (0.1444)  loss: 0.8023 (0.7977)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [54]  [1200/6250]  eta: 0:26:20  lr: 0.000059  grad: 0.1267 (0.1441)  loss: 0.7990 (0.7977)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [54]  [1300/6250]  eta: 0:25:32  lr: 0.000059  grad: 0.1329 (0.1438)  loss: 0.7943 (0.7977)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [54]  [1400/6250]  eta: 0:24:47  lr: 0.000059  grad: 0.1297 (0.1434)  loss: 0.8002 (0.7978)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [54]  [1500/6250]  eta: 0:24:18  lr: 0.000059  grad: 0.1282 (0.1437)  loss: 0.7994 (0.7978)  time: 0.4743  data: 0.1976  max mem: 26157
Train: [54]  [1600/6250]  eta: 0:23:37  lr: 0.000059  grad: 0.1249 (0.1438)  loss: 0.7940 (0.7976)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [54]  [1700/6250]  eta: 0:22:57  lr: 0.000059  grad: 0.1272 (0.1437)  loss: 0.7978 (0.7974)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [54]  [1800/6250]  eta: 0:22:19  lr: 0.000059  grad: 0.1276 (0.1440)  loss: 0.7967 (0.7973)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [54]  [1900/6250]  eta: 0:21:42  lr: 0.000059  grad: 0.1285 (0.1446)  loss: 0.7954 (0.7971)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [54]  [2000/6250]  eta: 0:21:05  lr: 0.000059  grad: 0.1300 (0.1456)  loss: 0.7902 (0.7969)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [54]  [2100/6250]  eta: 0:20:30  lr: 0.000059  grad: 0.1440 (0.1464)  loss: 0.7901 (0.7969)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [54]  [2200/6250]  eta: 0:19:56  lr: 0.000059  grad: 0.1283 (0.1462)  loss: 0.7936 (0.7967)  time: 0.2741  data: 0.0003  max mem: 26157
Train: [54]  [2300/6250]  eta: 0:19:22  lr: 0.000059  grad: 0.1339 (0.1464)  loss: 0.7904 (0.7966)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [54]  [2400/6250]  eta: 0:18:48  lr: 0.000059  grad: 0.1356 (0.1467)  loss: 0.7906 (0.7965)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [54]  [2500/6250]  eta: 0:18:16  lr: 0.000059  grad: 0.1221 (0.1472)  loss: 0.7928 (0.7963)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [54]  [2600/6250]  eta: 0:17:44  lr: 0.000059  grad: 0.1294 (0.1472)  loss: 0.7925 (0.7961)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [54]  [2700/6250]  eta: 0:17:12  lr: 0.000059  grad: 0.1284 (0.1471)  loss: 0.7951 (0.7960)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [54]  [2800/6250]  eta: 0:16:40  lr: 0.000058  grad: 0.1220 (0.1469)  loss: 0.8005 (0.7959)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [54]  [2900/6250]  eta: 0:16:09  lr: 0.000058  grad: 0.1310 (0.1474)  loss: 0.7951 (0.7959)  time: 0.2690  data: 0.0003  max mem: 26157
Train: [54]  [3000/6250]  eta: 0:15:38  lr: 0.000058  grad: 0.1290 (0.1475)  loss: 0.8004 (0.7959)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [54]  [3100/6250]  eta: 0:15:07  lr: 0.000058  grad: 0.1236 (0.1477)  loss: 0.7932 (0.7959)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [54]  [3200/6250]  eta: 0:14:36  lr: 0.000058  grad: 0.1240 (0.1476)  loss: 0.8000 (0.7959)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [54]  [3300/6250]  eta: 0:14:06  lr: 0.000058  grad: 0.1280 (0.1478)  loss: 0.7971 (0.7958)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [54]  [3400/6250]  eta: 0:13:36  lr: 0.000058  grad: 0.1231 (0.1477)  loss: 0.8043 (0.7959)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [54]  [3500/6250]  eta: 0:13:07  lr: 0.000058  grad: 0.1285 (0.1478)  loss: 0.7919 (0.7959)  time: 0.2768  data: 0.0002  max mem: 26157
Train: [54]  [3600/6250]  eta: 0:12:38  lr: 0.000058  grad: 0.1370 (0.1478)  loss: 0.7946 (0.7959)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [54]  [3700/6250]  eta: 0:12:08  lr: 0.000058  grad: 0.1294 (0.1478)  loss: 0.7963 (0.7959)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [54]  [3800/6250]  eta: 0:11:38  lr: 0.000058  grad: 0.1260 (0.1476)  loss: 0.7920 (0.7959)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [54]  [3900/6250]  eta: 0:11:09  lr: 0.000058  grad: 0.1244 (0.1476)  loss: 0.7999 (0.7959)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [54]  [4000/6250]  eta: 0:10:43  lr: 0.000058  grad: 0.1330 (0.1476)  loss: 0.7979 (0.7960)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [54]  [4100/6250]  eta: 0:10:13  lr: 0.000058  grad: 0.1312 (0.1477)  loss: 0.7886 (0.7960)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [54]  [4200/6250]  eta: 0:09:44  lr: 0.000058  grad: 0.1283 (0.1478)  loss: 0.7955 (0.7960)  time: 0.2763  data: 0.0002  max mem: 26157
Train: [54]  [4300/6250]  eta: 0:09:15  lr: 0.000058  grad: 0.1302 (0.1481)  loss: 0.7940 (0.7960)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [54]  [4400/6250]  eta: 0:08:46  lr: 0.000058  grad: 0.1449 (0.1499)  loss: 0.7968 (0.7960)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [54]  [4500/6250]  eta: 0:08:17  lr: 0.000058  grad: 0.1391 (0.1502)  loss: 0.7927 (0.7959)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [54]  [4600/6250]  eta: 0:07:48  lr: 0.000058  grad: 0.1317 (0.1504)  loss: 0.7992 (0.7959)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [54]  [4700/6250]  eta: 0:07:19  lr: 0.000058  grad: 0.1421 (0.1506)  loss: 0.7908 (0.7959)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [54]  [4800/6250]  eta: 0:06:50  lr: 0.000058  grad: 0.1398 (0.1509)  loss: 0.8047 (0.7960)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [54]  [4900/6250]  eta: 0:06:22  lr: 0.000058  grad: 0.1265 (0.1508)  loss: 0.8013 (0.7960)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [54]  [5000/6250]  eta: 0:05:53  lr: 0.000058  grad: 0.1252 (0.1512)  loss: 0.8000 (0.7960)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [54]  [5100/6250]  eta: 0:05:26  lr: 0.000058  grad: 0.1363 (0.1509)  loss: 0.7946 (0.7960)  time: 0.2730  data: 0.0002  max mem: 26157
Train: [54]  [5200/6250]  eta: 0:04:58  lr: 0.000058  grad: 0.1338 (0.1508)  loss: 0.7945 (0.7960)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [54]  [5300/6250]  eta: 0:04:29  lr: 0.000058  grad: 0.1421 (0.1509)  loss: 0.7985 (0.7959)  time: 0.3238  data: 0.0458  max mem: 26157
Train: [54]  [5400/6250]  eta: 0:04:01  lr: 0.000058  grad: 0.1346 (0.1510)  loss: 0.7954 (0.7959)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [54]  [5500/6250]  eta: 0:03:33  lr: 0.000058  grad: 0.1302 (0.1512)  loss: 0.7919 (0.7959)  time: 0.2712  data: 0.0001  max mem: 26157
Train: [54]  [5600/6250]  eta: 0:03:04  lr: 0.000058  grad: 0.1297 (0.1512)  loss: 0.7943 (0.7959)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [54]  [5700/6250]  eta: 0:02:35  lr: 0.000058  grad: 0.1343 (0.1517)  loss: 0.7987 (0.7959)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [54]  [5800/6250]  eta: 0:02:07  lr: 0.000057  grad: 0.1402 (0.1520)  loss: 0.7994 (0.7959)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [54]  [5900/6250]  eta: 0:01:39  lr: 0.000057  grad: 0.1311 (0.1519)  loss: 0.8052 (0.7960)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [54]  [6000/6250]  eta: 0:01:10  lr: 0.000057  grad: 0.1258 (0.1519)  loss: 0.7966 (0.7960)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [54]  [6100/6250]  eta: 0:00:42  lr: 0.000057  grad: 0.1281 (0.1519)  loss: 0.7971 (0.7960)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [54]  [6200/6250]  eta: 0:00:14  lr: 0.000057  grad: 0.1502 (0.1518)  loss: 0.7876 (0.7960)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [54]  [6249/6250]  eta: 0:00:00  lr: 0.000057  grad: 0.1353 (0.1519)  loss: 0.7933 (0.7960)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [54] Total time: 0:29:30 (0.2833 s / it)
Averaged stats: lr: 0.000057  grad: 0.1353 (0.1519)  loss: 0.7933 (0.7960)
Eval (hcp-train-subset): [54]  [ 0/62]  eta: 0:05:18  loss: 0.8347 (0.8347)  time: 5.1423  data: 5.0550  max mem: 26157
Eval (hcp-train-subset): [54]  [61/62]  eta: 0:00:00  loss: 0.8195 (0.8213)  time: 0.1208  data: 0.0359  max mem: 26157
Eval (hcp-train-subset): [54] Total time: 0:00:12 (0.2062 s / it)
Averaged stats (hcp-train-subset): loss: 0.8195 (0.8213)
Making plots (hcp-train-subset): example=60
Eval (hcp-val): [54]  [ 0/62]  eta: 0:03:28  loss: 0.8194 (0.8194)  time: 3.3557  data: 3.2471  max mem: 26157
Eval (hcp-val): [54]  [61/62]  eta: 0:00:00  loss: 0.8249 (0.8251)  time: 0.1282  data: 0.0436  max mem: 26157
Eval (hcp-val): [54] Total time: 0:00:13 (0.2155 s / it)
Averaged stats (hcp-val): loss: 0.8249 (0.8251)
Making plots (hcp-val): example=59
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [55]  [   0/6250]  eta: 8:40:06  lr: 0.000057  grad: 0.3015 (0.3015)  loss: 0.8035 (0.8035)  time: 4.9931  data: 4.6598  max mem: 26157
Train: [55]  [ 100/6250]  eta: 0:33:53  lr: 0.000057  grad: 0.1576 (0.2045)  loss: 0.7975 (0.8085)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [55]  [ 200/6250]  eta: 0:30:18  lr: 0.000057  grad: 0.1415 (0.1822)  loss: 0.7985 (0.8035)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [55]  [ 300/6250]  eta: 0:28:46  lr: 0.000057  grad: 0.1331 (0.1716)  loss: 0.8109 (0.8030)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [55]  [ 400/6250]  eta: 0:27:48  lr: 0.000057  grad: 0.1340 (0.1644)  loss: 0.7940 (0.8024)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [55]  [ 500/6250]  eta: 0:27:02  lr: 0.000057  grad: 0.1255 (0.1584)  loss: 0.8121 (0.8019)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [55]  [ 600/6250]  eta: 0:26:21  lr: 0.000057  grad: 0.1277 (0.1567)  loss: 0.7919 (0.8013)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [55]  [ 700/6250]  eta: 0:25:45  lr: 0.000057  grad: 0.1194 (0.1536)  loss: 0.7989 (0.8012)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [55]  [ 800/6250]  eta: 0:25:12  lr: 0.000057  grad: 0.1334 (0.1526)  loss: 0.8000 (0.8009)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [55]  [ 900/6250]  eta: 0:24:40  lr: 0.000057  grad: 0.1244 (0.1516)  loss: 0.8006 (0.8008)  time: 0.2704  data: 0.0003  max mem: 26157
Train: [55]  [1000/6250]  eta: 0:24:09  lr: 0.000057  grad: 0.1261 (0.1497)  loss: 0.8018 (0.8005)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [55]  [1100/6250]  eta: 0:23:38  lr: 0.000057  grad: 0.1229 (0.1480)  loss: 0.8016 (0.8005)  time: 0.2724  data: 0.0003  max mem: 26157
Train: [55]  [1200/6250]  eta: 0:23:09  lr: 0.000057  grad: 0.1238 (0.1470)  loss: 0.7958 (0.8004)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [55]  [1300/6250]  eta: 0:22:40  lr: 0.000057  grad: 0.1222 (0.1462)  loss: 0.8026 (0.8004)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [55]  [1400/6250]  eta: 0:22:29  lr: 0.000057  grad: 0.1282 (0.1458)  loss: 0.7960 (0.8002)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [55]  [1500/6250]  eta: 0:22:22  lr: 0.000057  grad: 0.1275 (0.1452)  loss: 0.7892 (0.8000)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [55]  [1600/6250]  eta: 0:21:51  lr: 0.000057  grad: 0.1259 (0.1453)  loss: 0.7904 (0.7998)  time: 0.2789  data: 0.0003  max mem: 26157
Train: [55]  [1700/6250]  eta: 0:21:48  lr: 0.000057  grad: 0.1343 (0.1457)  loss: 0.7944 (0.7995)  time: 0.3086  data: 0.0341  max mem: 26157
Train: [55]  [1800/6250]  eta: 0:21:15  lr: 0.000057  grad: 0.1237 (0.1455)  loss: 0.7985 (0.7992)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [55]  [1900/6250]  eta: 0:20:43  lr: 0.000057  grad: 0.1325 (0.1455)  loss: 0.7900 (0.7989)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [55]  [2000/6250]  eta: 0:20:11  lr: 0.000057  grad: 0.1306 (0.1459)  loss: 0.7905 (0.7985)  time: 0.2774  data: 0.0002  max mem: 26157
Train: [55]  [2100/6250]  eta: 0:19:51  lr: 0.000057  grad: 0.1256 (0.1459)  loss: 0.7901 (0.7982)  time: 0.5129  data: 0.2352  max mem: 26157
Train: [55]  [2200/6250]  eta: 0:19:32  lr: 0.000057  grad: 0.1246 (0.1461)  loss: 0.7886 (0.7980)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [55]  [2300/6250]  eta: 0:19:00  lr: 0.000057  grad: 0.1421 (0.1460)  loss: 0.7865 (0.7976)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [55]  [2400/6250]  eta: 0:18:28  lr: 0.000057  grad: 0.1308 (0.1458)  loss: 0.7925 (0.7975)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [55]  [2500/6250]  eta: 0:17:56  lr: 0.000057  grad: 0.1278 (0.1457)  loss: 0.8004 (0.7975)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [55]  [2600/6250]  eta: 0:17:25  lr: 0.000056  grad: 0.1282 (0.1454)  loss: 0.7980 (0.7975)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [55]  [2700/6250]  eta: 0:16:54  lr: 0.000056  grad: 0.1473 (0.1457)  loss: 0.7939 (0.7974)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [55]  [2800/6250]  eta: 0:16:23  lr: 0.000056  grad: 0.2013 (0.1482)  loss: 0.8000 (0.7974)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [55]  [2900/6250]  eta: 0:15:53  lr: 0.000056  grad: 0.1249 (0.1483)  loss: 0.7941 (0.7972)  time: 0.2718  data: 0.0002  max mem: 26157
Train: [55]  [3000/6250]  eta: 0:15:23  lr: 0.000056  grad: 0.1719 (0.1489)  loss: 0.7927 (0.7971)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [55]  [3100/6250]  eta: 0:14:53  lr: 0.000056  grad: 0.1653 (0.1502)  loss: 0.7806 (0.7968)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [55]  [3200/6250]  eta: 0:14:24  lr: 0.000056  grad: 0.1352 (0.1519)  loss: 0.7928 (0.7966)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [55]  [3300/6250]  eta: 0:14:02  lr: 0.000056  grad: 0.1550 (0.1533)  loss: 0.7925 (0.7964)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [55]  [3400/6250]  eta: 0:13:32  lr: 0.000056  grad: 0.1305 (0.1538)  loss: 0.7913 (0.7963)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [55]  [3500/6250]  eta: 0:13:02  lr: 0.000056  grad: 0.1324 (0.1535)  loss: 0.7939 (0.7962)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [55]  [3600/6250]  eta: 0:12:32  lr: 0.000056  grad: 0.1596 (0.1540)  loss: 0.7879 (0.7960)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [55]  [3700/6250]  eta: 0:12:03  lr: 0.000056  grad: 0.1428 (0.1546)  loss: 0.7959 (0.7958)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [55]  [3800/6250]  eta: 0:11:34  lr: 0.000056  grad: 0.1465 (0.1546)  loss: 0.7878 (0.7957)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [55]  [3900/6250]  eta: 0:11:04  lr: 0.000056  grad: 0.1375 (0.1547)  loss: 0.7970 (0.7957)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [55]  [4000/6250]  eta: 0:10:35  lr: 0.000056  grad: 0.1289 (0.1548)  loss: 0.7956 (0.7955)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [55]  [4100/6250]  eta: 0:10:06  lr: 0.000056  grad: 0.1354 (0.1548)  loss: 0.7949 (0.7955)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [55]  [4200/6250]  eta: 0:09:38  lr: 0.000056  grad: 0.1349 (0.1545)  loss: 0.7962 (0.7954)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [55]  [4300/6250]  eta: 0:09:12  lr: 0.000056  grad: 0.1313 (0.1543)  loss: 0.7910 (0.7954)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [55]  [4400/6250]  eta: 0:08:43  lr: 0.000056  grad: 0.1456 (0.1548)  loss: 0.7890 (0.7953)  time: 0.2738  data: 0.0002  max mem: 26157
Train: [55]  [4500/6250]  eta: 0:08:15  lr: 0.000056  grad: 0.1337 (0.1550)  loss: 0.7956 (0.7952)  time: 0.2755  data: 0.0002  max mem: 26157
Train: [55]  [4600/6250]  eta: 0:07:46  lr: 0.000056  grad: 0.1472 (0.1548)  loss: 0.7942 (0.7951)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [55]  [4700/6250]  eta: 0:07:18  lr: 0.000056  grad: 0.1326 (0.1545)  loss: 0.7902 (0.7950)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [55]  [4800/6250]  eta: 0:06:49  lr: 0.000056  grad: 0.1318 (0.1543)  loss: 0.7929 (0.7950)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [55]  [4900/6250]  eta: 0:06:20  lr: 0.000056  grad: 0.1392 (0.1541)  loss: 0.7843 (0.7949)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [55]  [5000/6250]  eta: 0:05:52  lr: 0.000056  grad: 0.1425 (0.1538)  loss: 0.7799 (0.7949)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [55]  [5100/6250]  eta: 0:05:24  lr: 0.000056  grad: 0.1303 (0.1537)  loss: 0.7939 (0.7948)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [55]  [5200/6250]  eta: 0:04:56  lr: 0.000056  grad: 0.1288 (0.1536)  loss: 0.7941 (0.7948)  time: 0.2734  data: 0.0002  max mem: 26157
Train: [55]  [5300/6250]  eta: 0:04:27  lr: 0.000056  grad: 0.1316 (0.1537)  loss: 0.7960 (0.7948)  time: 0.2731  data: 0.0002  max mem: 26157
Train: [55]  [5400/6250]  eta: 0:04:00  lr: 0.000056  grad: 0.1516 (0.1536)  loss: 0.7909 (0.7947)  time: 0.5075  data: 0.2238  max mem: 26157
Train: [55]  [5500/6250]  eta: 0:03:32  lr: 0.000056  grad: 0.1459 (0.1536)  loss: 0.7999 (0.7947)  time: 0.2761  data: 0.0002  max mem: 26157
Train: [55]  [5600/6250]  eta: 0:03:03  lr: 0.000055  grad: 0.1395 (0.1534)  loss: 0.7986 (0.7947)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [55]  [5700/6250]  eta: 0:02:36  lr: 0.000055  grad: 0.1348 (0.1537)  loss: 0.7966 (0.7947)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [55]  [5800/6250]  eta: 0:02:07  lr: 0.000055  grad: 0.1534 (0.1536)  loss: 0.7955 (0.7947)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [55]  [5900/6250]  eta: 0:01:39  lr: 0.000055  grad: 0.1382 (0.1536)  loss: 0.8025 (0.7948)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [55]  [6000/6250]  eta: 0:01:10  lr: 0.000055  grad: 0.1314 (0.1536)  loss: 0.8019 (0.7948)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [55]  [6100/6250]  eta: 0:00:42  lr: 0.000055  grad: 0.1358 (0.1536)  loss: 0.7974 (0.7949)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [55]  [6200/6250]  eta: 0:00:14  lr: 0.000055  grad: 0.1347 (0.1534)  loss: 0.7933 (0.7950)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [55]  [6249/6250]  eta: 0:00:00  lr: 0.000055  grad: 0.1325 (0.1534)  loss: 0.8038 (0.7950)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [55] Total time: 0:29:33 (0.2838 s / it)
Averaged stats: lr: 0.000055  grad: 0.1325 (0.1534)  loss: 0.8038 (0.7950)
Eval (hcp-train-subset): [55]  [ 0/62]  eta: 0:05:32  loss: 0.8294 (0.8294)  time: 5.3578  data: 5.2728  max mem: 26157
Eval (hcp-train-subset): [55]  [61/62]  eta: 0:00:00  loss: 0.8171 (0.8192)  time: 0.1394  data: 0.0567  max mem: 26157
Eval (hcp-train-subset): [55] Total time: 0:00:13 (0.2114 s / it)
Averaged stats (hcp-train-subset): loss: 0.8171 (0.8192)
Making plots (hcp-train-subset): example=42
Eval (hcp-val): [55]  [ 0/62]  eta: 0:03:37  loss: 0.8190 (0.8190)  time: 3.5055  data: 3.4009  max mem: 26157
Eval (hcp-val): [55]  [61/62]  eta: 0:00:00  loss: 0.8237 (0.8248)  time: 0.1394  data: 0.0542  max mem: 26157
Eval (hcp-val): [55] Total time: 0:00:13 (0.2179 s / it)
Averaged stats (hcp-val): loss: 0.8237 (0.8248)
Making plots (hcp-val): example=54
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [56]  [   0/6250]  eta: 9:42:35  lr: 0.000055  grad: 0.2590 (0.2590)  loss: 0.8266 (0.8266)  time: 5.5930  data: 5.2986  max mem: 26157
Train: [56]  [ 100/6250]  eta: 0:33:52  lr: 0.000055  grad: 0.1565 (0.1534)  loss: 0.8120 (0.8250)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [56]  [ 200/6250]  eta: 0:30:18  lr: 0.000055  grad: 0.1613 (0.1529)  loss: 0.7987 (0.8175)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [56]  [ 300/6250]  eta: 0:28:51  lr: 0.000055  grad: 0.1403 (0.1499)  loss: 0.7991 (0.8137)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [56]  [ 400/6250]  eta: 0:27:52  lr: 0.000055  grad: 0.1394 (0.1511)  loss: 0.8006 (0.8105)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [56]  [ 500/6250]  eta: 0:27:05  lr: 0.000055  grad: 0.1360 (0.1526)  loss: 0.7947 (0.8078)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [56]  [ 600/6250]  eta: 0:26:25  lr: 0.000055  grad: 0.1390 (0.1519)  loss: 0.7921 (0.8057)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [56]  [ 700/6250]  eta: 0:26:08  lr: 0.000055  grad: 0.1524 (0.1515)  loss: 0.7985 (0.8040)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [56]  [ 800/6250]  eta: 0:25:32  lr: 0.000055  grad: 0.1206 (0.1508)  loss: 0.7986 (0.8034)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [56]  [ 900/6250]  eta: 0:25:29  lr: 0.000055  grad: 0.1385 (0.1495)  loss: 0.8009 (0.8029)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [56]  [1000/6250]  eta: 0:25:35  lr: 0.000055  grad: 0.1339 (0.1500)  loss: 0.7940 (0.8024)  time: 0.2744  data: 0.0002  max mem: 26157
Train: [56]  [1100/6250]  eta: 0:24:56  lr: 0.000055  grad: 0.1347 (0.1499)  loss: 0.7900 (0.8019)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [56]  [1200/6250]  eta: 0:24:18  lr: 0.000055  grad: 0.1271 (0.1492)  loss: 0.7954 (0.8016)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [56]  [1300/6250]  eta: 0:23:41  lr: 0.000055  grad: 0.1313 (0.1490)  loss: 0.7967 (0.8013)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [56]  [1400/6250]  eta: 0:23:06  lr: 0.000055  grad: 0.1374 (0.1486)  loss: 0.7905 (0.8008)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [56]  [1500/6250]  eta: 0:22:32  lr: 0.000055  grad: 0.1338 (0.1486)  loss: 0.7902 (0.8004)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [56]  [1600/6250]  eta: 0:22:00  lr: 0.000055  grad: 0.1244 (0.1481)  loss: 0.7953 (0.8001)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [56]  [1700/6250]  eta: 0:21:28  lr: 0.000055  grad: 0.1299 (0.1491)  loss: 0.7938 (0.7998)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [56]  [1800/6250]  eta: 0:20:56  lr: 0.000055  grad: 0.1261 (0.1496)  loss: 0.7951 (0.7994)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [56]  [1900/6250]  eta: 0:20:25  lr: 0.000055  grad: 0.1386 (0.1495)  loss: 0.7874 (0.7991)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [56]  [2000/6250]  eta: 0:19:54  lr: 0.000055  grad: 0.1273 (0.1490)  loss: 0.7934 (0.7988)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [56]  [2100/6250]  eta: 0:19:26  lr: 0.000055  grad: 0.1318 (0.1491)  loss: 0.7958 (0.7985)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [56]  [2200/6250]  eta: 0:18:56  lr: 0.000055  grad: 0.1371 (0.1497)  loss: 0.7878 (0.7982)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [56]  [2300/6250]  eta: 0:18:26  lr: 0.000055  grad: 0.1363 (0.1491)  loss: 0.7952 (0.7980)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [56]  [2400/6250]  eta: 0:17:56  lr: 0.000054  grad: 0.1330 (0.1490)  loss: 0.7948 (0.7977)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [56]  [2500/6250]  eta: 0:17:26  lr: 0.000054  grad: 0.1522 (0.1495)  loss: 0.7981 (0.7974)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [56]  [2600/6250]  eta: 0:16:57  lr: 0.000054  grad: 0.1342 (0.1495)  loss: 0.7927 (0.7972)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [56]  [2700/6250]  eta: 0:16:28  lr: 0.000054  grad: 0.1440 (0.1499)  loss: 0.7985 (0.7971)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [56]  [2800/6250]  eta: 0:15:59  lr: 0.000054  grad: 0.1317 (0.1500)  loss: 0.8008 (0.7970)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [56]  [2900/6250]  eta: 0:15:30  lr: 0.000054  grad: 0.1376 (0.1502)  loss: 0.7911 (0.7969)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [56]  [3000/6250]  eta: 0:15:02  lr: 0.000054  grad: 0.1486 (0.1505)  loss: 0.7920 (0.7967)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [56]  [3100/6250]  eta: 0:14:33  lr: 0.000054  grad: 0.1469 (0.1506)  loss: 0.7884 (0.7967)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [56]  [3200/6250]  eta: 0:14:07  lr: 0.000054  grad: 0.1285 (0.1503)  loss: 0.7969 (0.7965)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [56]  [3300/6250]  eta: 0:13:40  lr: 0.000054  grad: 0.1353 (0.1500)  loss: 0.7914 (0.7964)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [56]  [3400/6250]  eta: 0:13:15  lr: 0.000054  grad: 0.1362 (0.1498)  loss: 0.7941 (0.7963)  time: 0.4716  data: 0.1993  max mem: 26157
Train: [56]  [3500/6250]  eta: 0:12:46  lr: 0.000054  grad: 0.1323 (0.1502)  loss: 0.7985 (0.7962)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [56]  [3600/6250]  eta: 0:12:26  lr: 0.000054  grad: 0.1464 (0.1503)  loss: 0.7906 (0.7961)  time: 0.4573  data: 0.1788  max mem: 26157
Train: [56]  [3700/6250]  eta: 0:11:57  lr: 0.000054  grad: 0.1355 (0.1508)  loss: 0.7914 (0.7959)  time: 0.2846  data: 0.0003  max mem: 26157
Train: [56]  [3800/6250]  eta: 0:11:28  lr: 0.000054  grad: 0.1483 (0.1510)  loss: 0.7904 (0.7958)  time: 0.2773  data: 0.0003  max mem: 26157
Train: [56]  [3900/6250]  eta: 0:11:00  lr: 0.000054  grad: 0.1356 (0.1511)  loss: 0.7934 (0.7957)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [56]  [4000/6250]  eta: 0:10:32  lr: 0.000054  grad: 0.1343 (0.1510)  loss: 0.7946 (0.7956)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [56]  [4100/6250]  eta: 0:10:03  lr: 0.000054  grad: 0.1378 (0.1513)  loss: 0.7852 (0.7955)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [56]  [4200/6250]  eta: 0:09:35  lr: 0.000054  grad: 0.1472 (0.1516)  loss: 0.7913 (0.7954)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [56]  [4300/6250]  eta: 0:09:06  lr: 0.000054  grad: 0.1349 (0.1517)  loss: 0.7915 (0.7952)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [56]  [4400/6250]  eta: 0:08:38  lr: 0.000054  grad: 0.1307 (0.1517)  loss: 0.7880 (0.7951)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [56]  [4500/6250]  eta: 0:08:09  lr: 0.000054  grad: 0.1296 (0.1518)  loss: 0.7904 (0.7949)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [56]  [4600/6250]  eta: 0:07:41  lr: 0.000054  grad: 0.1344 (0.1517)  loss: 0.7857 (0.7948)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [56]  [4700/6250]  eta: 0:07:13  lr: 0.000054  grad: 0.1387 (0.1519)  loss: 0.7854 (0.7945)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [56]  [4800/6250]  eta: 0:06:45  lr: 0.000054  grad: 0.1413 (0.1520)  loss: 0.7846 (0.7944)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [56]  [4900/6250]  eta: 0:06:17  lr: 0.000054  grad: 0.1454 (0.1520)  loss: 0.7856 (0.7943)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [56]  [5000/6250]  eta: 0:05:50  lr: 0.000054  grad: 0.1340 (0.1520)  loss: 0.7911 (0.7942)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [56]  [5100/6250]  eta: 0:05:21  lr: 0.000054  grad: 0.1382 (0.1521)  loss: 0.7794 (0.7940)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [56]  [5200/6250]  eta: 0:04:53  lr: 0.000054  grad: 0.1340 (0.1522)  loss: 0.7881 (0.7939)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [56]  [5300/6250]  eta: 0:04:25  lr: 0.000054  grad: 0.1370 (0.1525)  loss: 0.7882 (0.7938)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [56]  [5400/6250]  eta: 0:03:57  lr: 0.000054  grad: 0.1406 (0.1525)  loss: 0.7846 (0.7937)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [56]  [5500/6250]  eta: 0:03:29  lr: 0.000053  grad: 0.1379 (0.1526)  loss: 0.7938 (0.7937)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [56]  [5600/6250]  eta: 0:03:01  lr: 0.000053  grad: 0.1380 (0.1530)  loss: 0.7878 (0.7936)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [56]  [5700/6250]  eta: 0:02:33  lr: 0.000053  grad: 0.1382 (0.1528)  loss: 0.7864 (0.7936)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [56]  [5800/6250]  eta: 0:02:05  lr: 0.000053  grad: 0.1314 (0.1528)  loss: 0.7889 (0.7935)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [56]  [5900/6250]  eta: 0:01:37  lr: 0.000053  grad: 0.1426 (0.1530)  loss: 0.7916 (0.7935)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [56]  [6000/6250]  eta: 0:01:09  lr: 0.000053  grad: 0.1313 (0.1530)  loss: 0.7922 (0.7935)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [56]  [6100/6250]  eta: 0:00:41  lr: 0.000053  grad: 0.1247 (0.1530)  loss: 0.7917 (0.7935)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [56]  [6200/6250]  eta: 0:00:13  lr: 0.000053  grad: 0.1482 (0.1532)  loss: 0.7875 (0.7934)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [56]  [6249/6250]  eta: 0:00:00  lr: 0.000053  grad: 0.1355 (0.1532)  loss: 0.7942 (0.7934)  time: 0.2773  data: 0.0002  max mem: 26157
Train: [56] Total time: 0:29:05 (0.2793 s / it)
Averaged stats: lr: 0.000053  grad: 0.1355 (0.1532)  loss: 0.7942 (0.7934)
Eval (hcp-train-subset): [56]  [ 0/62]  eta: 0:04:16  loss: 0.8312 (0.8312)  time: 4.1324  data: 4.0084  max mem: 26157
Eval (hcp-train-subset): [56]  [61/62]  eta: 0:00:00  loss: 0.8176 (0.8183)  time: 0.1345  data: 0.0499  max mem: 26157
Eval (hcp-train-subset): [56] Total time: 0:00:13 (0.2242 s / it)
Averaged stats (hcp-train-subset): loss: 0.8176 (0.8183)
Making plots (hcp-train-subset): example=45
Eval (hcp-val): [56]  [ 0/62]  eta: 0:03:48  loss: 0.8197 (0.8197)  time: 3.6789  data: 3.5567  max mem: 26157
Eval (hcp-val): [56]  [61/62]  eta: 0:00:00  loss: 0.8239 (0.8246)  time: 0.1460  data: 0.0630  max mem: 26157
Eval (hcp-val): [56] Total time: 0:00:13 (0.2137 s / it)
Averaged stats (hcp-val): loss: 0.8239 (0.8246)
Making plots (hcp-val): example=16
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [57]  [   0/6250]  eta: 10:33:25  lr: 0.000053  grad: 0.3559 (0.3559)  loss: 0.7869 (0.7869)  time: 6.0809  data: 5.8051  max mem: 26157
Train: [57]  [ 100/6250]  eta: 0:34:10  lr: 0.000053  grad: 0.1429 (0.1735)  loss: 0.8076 (0.8157)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [57]  [ 200/6250]  eta: 0:30:27  lr: 0.000053  grad: 0.1447 (0.1651)  loss: 0.8013 (0.8094)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [57]  [ 300/6250]  eta: 0:28:56  lr: 0.000053  grad: 0.1324 (0.1586)  loss: 0.8039 (0.8084)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [57]  [ 400/6250]  eta: 0:27:56  lr: 0.000053  grad: 0.1404 (0.1558)  loss: 0.7973 (0.8073)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [57]  [ 500/6250]  eta: 0:27:13  lr: 0.000053  grad: 0.1292 (0.1553)  loss: 0.7994 (0.8067)  time: 0.2780  data: 0.0003  max mem: 26157
Train: [57]  [ 600/6250]  eta: 0:26:35  lr: 0.000053  grad: 0.1511 (0.1563)  loss: 0.7950 (0.8058)  time: 0.2712  data: 0.0003  max mem: 26157
Train: [57]  [ 700/6250]  eta: 0:27:19  lr: 0.000053  grad: 0.1593 (0.1639)  loss: 0.8017 (0.8050)  time: 0.2772  data: 0.0002  max mem: 26157
Train: [57]  [ 800/6250]  eta: 0:26:36  lr: 0.000053  grad: 0.1675 (0.1735)  loss: 0.7973 (0.8043)  time: 0.2772  data: 0.0003  max mem: 26157
Train: [57]  [ 900/6250]  eta: 0:26:50  lr: 0.000053  grad: 0.1388 (0.1751)  loss: 0.8040 (0.8038)  time: 0.7217  data: 0.4471  max mem: 26157
Train: [57]  [1000/6250]  eta: 0:26:05  lr: 0.000053  grad: 0.1232 (0.1716)  loss: 0.7998 (0.8032)  time: 0.2755  data: 0.0002  max mem: 26157
Train: [57]  [1100/6250]  eta: 0:25:22  lr: 0.000053  grad: 0.1305 (0.1688)  loss: 0.8015 (0.8028)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [57]  [1200/6250]  eta: 0:25:17  lr: 0.000053  grad: 0.1328 (0.1664)  loss: 0.7948 (0.8024)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [57]  [1300/6250]  eta: 0:24:36  lr: 0.000053  grad: 0.1289 (0.1645)  loss: 0.7923 (0.8017)  time: 0.2774  data: 0.0002  max mem: 26157
Train: [57]  [1400/6250]  eta: 0:24:09  lr: 0.000053  grad: 0.1363 (0.1636)  loss: 0.7961 (0.8011)  time: 0.2695  data: 0.0003  max mem: 26157
Train: [57]  [1500/6250]  eta: 0:23:30  lr: 0.000053  grad: 0.1427 (0.1630)  loss: 0.7949 (0.8005)  time: 0.2706  data: 0.0007  max mem: 26157
Train: [57]  [1600/6250]  eta: 0:22:55  lr: 0.000053  grad: 0.1471 (0.1640)  loss: 0.7887 (0.7999)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [57]  [1700/6250]  eta: 0:22:20  lr: 0.000053  grad: 0.1301 (0.1624)  loss: 0.7909 (0.7996)  time: 0.2742  data: 0.0002  max mem: 26157
Train: [57]  [1800/6250]  eta: 0:21:45  lr: 0.000053  grad: 0.1354 (0.1625)  loss: 0.7886 (0.7991)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [57]  [1900/6250]  eta: 0:21:22  lr: 0.000053  grad: 0.1399 (0.1614)  loss: 0.7903 (0.7987)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [57]  [2000/6250]  eta: 0:20:47  lr: 0.000053  grad: 0.1322 (0.1611)  loss: 0.7875 (0.7984)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [57]  [2100/6250]  eta: 0:20:22  lr: 0.000053  grad: 0.1435 (0.1611)  loss: 0.7903 (0.7979)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [57]  [2200/6250]  eta: 0:19:55  lr: 0.000053  grad: 0.1512 (0.1621)  loss: 0.7929 (0.7975)  time: 0.4588  data: 0.1744  max mem: 26157
Train: [57]  [2300/6250]  eta: 0:19:21  lr: 0.000052  grad: 0.1542 (0.1624)  loss: 0.7859 (0.7972)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [57]  [2400/6250]  eta: 0:18:48  lr: 0.000052  grad: 0.1368 (0.1618)  loss: 0.7943 (0.7970)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [57]  [2500/6250]  eta: 0:18:15  lr: 0.000052  grad: 0.1347 (0.1614)  loss: 0.7910 (0.7967)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [57]  [2600/6250]  eta: 0:17:54  lr: 0.000052  grad: 0.1435 (0.1612)  loss: 0.7932 (0.7964)  time: 0.6714  data: 0.4000  max mem: 26157
Train: [57]  [2700/6250]  eta: 0:17:21  lr: 0.000052  grad: 0.1404 (0.1609)  loss: 0.7891 (0.7961)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [57]  [2800/6250]  eta: 0:16:54  lr: 0.000052  grad: 0.1365 (0.1609)  loss: 0.7881 (0.7959)  time: 0.4649  data: 0.1936  max mem: 26157
Train: [57]  [2900/6250]  eta: 0:16:27  lr: 0.000052  grad: 0.1450 (0.1616)  loss: 0.7927 (0.7956)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [57]  [3000/6250]  eta: 0:15:55  lr: 0.000052  grad: 0.1369 (0.1620)  loss: 0.7874 (0.7955)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [57]  [3100/6250]  eta: 0:15:28  lr: 0.000052  grad: 0.1483 (0.1620)  loss: 0.7795 (0.7953)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [57]  [3200/6250]  eta: 0:14:57  lr: 0.000052  grad: 0.1371 (0.1620)  loss: 0.7969 (0.7952)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [57]  [3300/6250]  eta: 0:14:32  lr: 0.000052  grad: 0.1509 (0.1620)  loss: 0.7958 (0.7952)  time: 0.2716  data: 0.0001  max mem: 26157
Train: [57]  [3400/6250]  eta: 0:14:00  lr: 0.000052  grad: 0.1396 (0.1616)  loss: 0.7946 (0.7951)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [57]  [3500/6250]  eta: 0:13:29  lr: 0.000052  grad: 0.1364 (0.1613)  loss: 0.7932 (0.7951)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [57]  [3600/6250]  eta: 0:12:57  lr: 0.000052  grad: 0.1474 (0.1615)  loss: 0.7980 (0.7949)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [57]  [3700/6250]  eta: 0:12:26  lr: 0.000052  grad: 0.1482 (0.1618)  loss: 0.7833 (0.7948)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [57]  [3800/6250]  eta: 0:11:56  lr: 0.000052  grad: 0.1403 (0.1617)  loss: 0.7905 (0.7947)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [57]  [3900/6250]  eta: 0:11:25  lr: 0.000052  grad: 0.1383 (0.1615)  loss: 0.7952 (0.7945)  time: 0.2690  data: 0.0003  max mem: 26157
Train: [57]  [4000/6250]  eta: 0:10:55  lr: 0.000052  grad: 0.1404 (0.1614)  loss: 0.7911 (0.7944)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [57]  [4100/6250]  eta: 0:10:24  lr: 0.000052  grad: 0.1431 (0.1616)  loss: 0.7931 (0.7943)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [57]  [4200/6250]  eta: 0:09:54  lr: 0.000052  grad: 0.1534 (0.1620)  loss: 0.7886 (0.7942)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [57]  [4300/6250]  eta: 0:09:24  lr: 0.000052  grad: 0.1432 (0.1618)  loss: 0.7918 (0.7941)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [57]  [4400/6250]  eta: 0:08:54  lr: 0.000052  grad: 0.1419 (0.1618)  loss: 0.7859 (0.7940)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [57]  [4500/6250]  eta: 0:08:25  lr: 0.000052  grad: 0.1394 (0.1615)  loss: 0.7872 (0.7939)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [57]  [4600/6250]  eta: 0:07:55  lr: 0.000052  grad: 0.1506 (0.1615)  loss: 0.7986 (0.7938)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [57]  [4700/6250]  eta: 0:07:26  lr: 0.000052  grad: 0.1387 (0.1616)  loss: 0.7928 (0.7938)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [57]  [4800/6250]  eta: 0:06:57  lr: 0.000052  grad: 0.1396 (0.1617)  loss: 0.7927 (0.7938)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [57]  [4900/6250]  eta: 0:06:27  lr: 0.000052  grad: 0.1352 (0.1617)  loss: 0.7997 (0.7938)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [57]  [5000/6250]  eta: 0:05:58  lr: 0.000052  grad: 0.1250 (0.1615)  loss: 0.7887 (0.7938)  time: 0.2727  data: 0.0002  max mem: 26157
Train: [57]  [5100/6250]  eta: 0:05:29  lr: 0.000052  grad: 0.1339 (0.1616)  loss: 0.7979 (0.7938)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [57]  [5200/6250]  eta: 0:05:00  lr: 0.000052  grad: 0.1428 (0.1616)  loss: 0.7936 (0.7939)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [57]  [5300/6250]  eta: 0:04:33  lr: 0.000052  grad: 0.1430 (0.1618)  loss: 0.7956 (0.7939)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [57]  [5400/6250]  eta: 0:04:04  lr: 0.000051  grad: 0.1423 (0.1622)  loss: 0.7950 (0.7939)  time: 0.4171  data: 0.1431  max mem: 26157
Train: [57]  [5500/6250]  eta: 0:03:35  lr: 0.000051  grad: 0.1501 (0.1623)  loss: 0.7919 (0.7939)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [57]  [5600/6250]  eta: 0:03:07  lr: 0.000051  grad: 0.1464 (0.1625)  loss: 0.7981 (0.7938)  time: 0.2745  data: 0.0002  max mem: 26157
Train: [57]  [5700/6250]  eta: 0:02:38  lr: 0.000051  grad: 0.1410 (0.1627)  loss: 0.7995 (0.7938)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [57]  [5800/6250]  eta: 0:02:09  lr: 0.000051  grad: 0.1338 (0.1626)  loss: 0.7931 (0.7938)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [57]  [5900/6250]  eta: 0:01:40  lr: 0.000051  grad: 0.1456 (0.1625)  loss: 0.7915 (0.7938)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [57]  [6000/6250]  eta: 0:01:11  lr: 0.000051  grad: 0.1314 (0.1626)  loss: 0.8007 (0.7939)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [57]  [6100/6250]  eta: 0:00:43  lr: 0.000051  grad: 0.1503 (0.1627)  loss: 0.7921 (0.7939)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [57]  [6200/6250]  eta: 0:00:14  lr: 0.000051  grad: 0.1563 (0.1630)  loss: 0.7937 (0.7939)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [57]  [6249/6250]  eta: 0:00:00  lr: 0.000051  grad: 0.1417 (0.1630)  loss: 0.7903 (0.7939)  time: 0.2762  data: 0.0002  max mem: 26157
Train: [57] Total time: 0:30:02 (0.2884 s / it)
Averaged stats: lr: 0.000051  grad: 0.1417 (0.1630)  loss: 0.7903 (0.7939)
Eval (hcp-train-subset): [57]  [ 0/62]  eta: 0:04:59  loss: 0.8244 (0.8244)  time: 4.8374  data: 4.7537  max mem: 26157
Eval (hcp-train-subset): [57]  [61/62]  eta: 0:00:00  loss: 0.8194 (0.8204)  time: 0.1321  data: 0.0493  max mem: 26157
Eval (hcp-train-subset): [57] Total time: 0:00:13 (0.2155 s / it)
Averaged stats (hcp-train-subset): loss: 0.8194 (0.8204)
Making plots (hcp-train-subset): example=7
Eval (hcp-val): [57]  [ 0/62]  eta: 0:05:26  loss: 0.8221 (0.8221)  time: 5.2623  data: 5.1790  max mem: 26157
Eval (hcp-val): [57]  [61/62]  eta: 0:00:00  loss: 0.8241 (0.8257)  time: 0.1347  data: 0.0495  max mem: 26157
Eval (hcp-val): [57] Total time: 0:00:13 (0.2111 s / it)
Averaged stats (hcp-val): loss: 0.8241 (0.8257)
Making plots (hcp-val): example=51
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [58]  [   0/6250]  eta: 9:27:26  lr: 0.000051  grad: 0.1346 (0.1346)  loss: 0.8441 (0.8441)  time: 5.4475  data: 5.0757  max mem: 26157
Train: [58]  [ 100/6250]  eta: 0:34:00  lr: 0.000051  grad: 0.1443 (0.1938)  loss: 0.8066 (0.8011)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [58]  [ 200/6250]  eta: 0:30:16  lr: 0.000051  grad: 0.1347 (0.1771)  loss: 0.8015 (0.7991)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [58]  [ 300/6250]  eta: 0:28:46  lr: 0.000051  grad: 0.1323 (0.1658)  loss: 0.7982 (0.7994)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [58]  [ 400/6250]  eta: 0:27:48  lr: 0.000051  grad: 0.1282 (0.1609)  loss: 0.8018 (0.8005)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [58]  [ 500/6250]  eta: 0:27:01  lr: 0.000051  grad: 0.1324 (0.1593)  loss: 0.8112 (0.8016)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [58]  [ 600/6250]  eta: 0:26:20  lr: 0.000051  grad: 0.1517 (0.1618)  loss: 0.8105 (0.8021)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [58]  [ 700/6250]  eta: 0:25:43  lr: 0.000051  grad: 0.1488 (0.1607)  loss: 0.8010 (0.8017)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [58]  [ 800/6250]  eta: 0:25:10  lr: 0.000051  grad: 0.1399 (0.1611)  loss: 0.7991 (0.8014)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [58]  [ 900/6250]  eta: 0:24:38  lr: 0.000051  grad: 0.1382 (0.1607)  loss: 0.8050 (0.8010)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [58]  [1000/6250]  eta: 0:24:07  lr: 0.000051  grad: 0.1467 (0.1593)  loss: 0.7962 (0.8008)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [58]  [1100/6250]  eta: 0:23:38  lr: 0.000051  grad: 0.1402 (0.1587)  loss: 0.7885 (0.8005)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [58]  [1200/6250]  eta: 0:23:08  lr: 0.000051  grad: 0.1271 (0.1583)  loss: 0.7948 (0.7999)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [58]  [1300/6250]  eta: 0:22:39  lr: 0.000051  grad: 0.1404 (0.1571)  loss: 0.8003 (0.7995)  time: 0.2732  data: 0.0002  max mem: 26157
Train: [58]  [1400/6250]  eta: 0:22:10  lr: 0.000051  grad: 0.1324 (0.1575)  loss: 0.7940 (0.7990)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [58]  [1500/6250]  eta: 0:21:55  lr: 0.000051  grad: 0.1350 (0.1578)  loss: 0.7918 (0.7985)  time: 0.4789  data: 0.2005  max mem: 26157
Train: [58]  [1600/6250]  eta: 0:22:05  lr: 0.000051  grad: 0.1569 (0.1580)  loss: 0.7990 (0.7981)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [58]  [1700/6250]  eta: 0:21:33  lr: 0.000051  grad: 0.1306 (0.1574)  loss: 0.7975 (0.7978)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [58]  [1800/6250]  eta: 0:21:17  lr: 0.000051  grad: 0.1504 (0.1577)  loss: 0.7916 (0.7974)  time: 0.5928  data: 0.3163  max mem: 26157
Train: [58]  [1900/6250]  eta: 0:20:47  lr: 0.000051  grad: 0.1367 (0.1576)  loss: 0.7942 (0.7969)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [58]  [2000/6250]  eta: 0:20:15  lr: 0.000051  grad: 0.1393 (0.1574)  loss: 0.7938 (0.7966)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [58]  [2100/6250]  eta: 0:19:43  lr: 0.000051  grad: 0.1453 (0.1572)  loss: 0.7891 (0.7965)  time: 0.2702  data: 0.0001  max mem: 26157
Train: [58]  [2200/6250]  eta: 0:19:15  lr: 0.000050  grad: 0.1440 (0.1579)  loss: 0.7949 (0.7963)  time: 0.2768  data: 0.0002  max mem: 26157
Train: [58]  [2300/6250]  eta: 0:18:45  lr: 0.000050  grad: 0.1420 (0.1577)  loss: 0.7978 (0.7961)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [58]  [2400/6250]  eta: 0:18:14  lr: 0.000050  grad: 0.1474 (0.1574)  loss: 0.7907 (0.7959)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [58]  [2500/6250]  eta: 0:17:44  lr: 0.000050  grad: 0.1376 (0.1579)  loss: 0.7883 (0.7957)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [58]  [2600/6250]  eta: 0:17:13  lr: 0.000050  grad: 0.1339 (0.1576)  loss: 0.7932 (0.7954)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [58]  [2700/6250]  eta: 0:16:43  lr: 0.000050  grad: 0.1399 (0.1580)  loss: 0.7879 (0.7953)  time: 0.2687  data: 0.0003  max mem: 26157
Train: [58]  [2800/6250]  eta: 0:16:23  lr: 0.000050  grad: 0.1392 (0.1577)  loss: 0.7915 (0.7952)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [58]  [2900/6250]  eta: 0:15:52  lr: 0.000050  grad: 0.1420 (0.1582)  loss: 0.7858 (0.7950)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [58]  [3000/6250]  eta: 0:15:23  lr: 0.000050  grad: 0.1477 (0.1591)  loss: 0.7940 (0.7948)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [58]  [3100/6250]  eta: 0:15:02  lr: 0.000050  grad: 0.1582 (0.1602)  loss: 0.7885 (0.7945)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [58]  [3200/6250]  eta: 0:14:33  lr: 0.000050  grad: 0.1589 (0.1608)  loss: 0.7902 (0.7943)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [58]  [3300/6250]  eta: 0:14:02  lr: 0.000050  grad: 0.1426 (0.1610)  loss: 0.7852 (0.7941)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [58]  [3400/6250]  eta: 0:13:39  lr: 0.000050  grad: 0.1561 (0.1612)  loss: 0.7866 (0.7939)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [58]  [3500/6250]  eta: 0:13:09  lr: 0.000050  grad: 0.1617 (0.1614)  loss: 0.7822 (0.7937)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [58]  [3600/6250]  eta: 0:12:40  lr: 0.000050  grad: 0.1515 (0.1622)  loss: 0.7943 (0.7935)  time: 0.3553  data: 0.0828  max mem: 26157
Train: [58]  [3700/6250]  eta: 0:12:12  lr: 0.000050  grad: 0.1650 (0.1634)  loss: 0.7870 (0.7932)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [58]  [3800/6250]  eta: 0:11:44  lr: 0.000050  grad: 0.1549 (0.1641)  loss: 0.7862 (0.7931)  time: 0.4108  data: 0.1334  max mem: 26157
Train: [58]  [3900/6250]  eta: 0:11:14  lr: 0.000050  grad: 0.1505 (0.1643)  loss: 0.7854 (0.7929)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [58]  [4000/6250]  eta: 0:10:47  lr: 0.000050  grad: 0.1536 (0.1643)  loss: 0.7812 (0.7928)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [58]  [4100/6250]  eta: 0:10:21  lr: 0.000050  grad: 0.1399 (0.1644)  loss: 0.7894 (0.7926)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [58]  [4200/6250]  eta: 0:09:51  lr: 0.000050  grad: 0.1400 (0.1643)  loss: 0.7851 (0.7925)  time: 0.2709  data: 0.0003  max mem: 26157
Train: [58]  [4300/6250]  eta: 0:09:21  lr: 0.000050  grad: 0.1392 (0.1641)  loss: 0.7877 (0.7924)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [58]  [4400/6250]  eta: 0:08:53  lr: 0.000050  grad: 0.1372 (0.1641)  loss: 0.7906 (0.7922)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [58]  [4500/6250]  eta: 0:08:24  lr: 0.000050  grad: 0.1527 (0.1641)  loss: 0.7840 (0.7921)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [58]  [4600/6250]  eta: 0:07:54  lr: 0.000050  grad: 0.1423 (0.1638)  loss: 0.7942 (0.7921)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [58]  [4700/6250]  eta: 0:07:26  lr: 0.000050  grad: 0.1416 (0.1641)  loss: 0.7861 (0.7921)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [58]  [4800/6250]  eta: 0:07:00  lr: 0.000050  grad: 0.1330 (0.1647)  loss: 0.7961 (0.7921)  time: 0.2729  data: 0.0002  max mem: 26157
Train: [58]  [4900/6250]  eta: 0:06:30  lr: 0.000050  grad: 0.1526 (0.1646)  loss: 0.7855 (0.7920)  time: 0.3112  data: 0.0361  max mem: 26157
Train: [58]  [5000/6250]  eta: 0:06:03  lr: 0.000050  grad: 0.1402 (0.1646)  loss: 0.7865 (0.7920)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [58]  [5100/6250]  eta: 0:05:33  lr: 0.000050  grad: 0.1336 (0.1643)  loss: 0.7946 (0.7919)  time: 0.2736  data: 0.0002  max mem: 26157
Train: [58]  [5200/6250]  eta: 0:05:04  lr: 0.000050  grad: 0.1506 (0.1642)  loss: 0.7906 (0.7919)  time: 0.2952  data: 0.0243  max mem: 26157
Train: [58]  [5300/6250]  eta: 0:04:35  lr: 0.000049  grad: 0.1436 (0.1642)  loss: 0.7888 (0.7918)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [58]  [5400/6250]  eta: 0:04:07  lr: 0.000049  grad: 0.1479 (0.1645)  loss: 0.7819 (0.7917)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [58]  [5500/6250]  eta: 0:03:37  lr: 0.000049  grad: 0.1549 (0.1644)  loss: 0.7786 (0.7916)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [58]  [5600/6250]  eta: 0:03:08  lr: 0.000049  grad: 0.1640 (0.1647)  loss: 0.7804 (0.7915)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [58]  [5700/6250]  eta: 0:02:39  lr: 0.000049  grad: 0.1530 (0.1647)  loss: 0.7828 (0.7913)  time: 0.2710  data: 0.0001  max mem: 26157
Train: [58]  [5800/6250]  eta: 0:02:10  lr: 0.000049  grad: 0.1471 (0.1651)  loss: 0.7954 (0.7913)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [58]  [5900/6250]  eta: 0:01:41  lr: 0.000049  grad: 0.1414 (0.1650)  loss: 0.7964 (0.7912)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [58]  [6000/6250]  eta: 0:01:12  lr: 0.000049  grad: 0.1489 (0.1652)  loss: 0.7857 (0.7912)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [58]  [6100/6250]  eta: 0:00:43  lr: 0.000049  grad: 0.1541 (0.1656)  loss: 0.7838 (0.7911)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [58]  [6200/6250]  eta: 0:00:14  lr: 0.000049  grad: 0.1394 (0.1657)  loss: 0.7942 (0.7910)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [58]  [6249/6250]  eta: 0:00:00  lr: 0.000049  grad: 0.1490 (0.1658)  loss: 0.7916 (0.7910)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [58] Total time: 0:30:08 (0.2894 s / it)
Averaged stats: lr: 0.000049  grad: 0.1490 (0.1658)  loss: 0.7916 (0.7910)
Eval (hcp-train-subset): [58]  [ 0/62]  eta: 0:03:21  loss: 0.8289 (0.8289)  time: 3.2461  data: 3.1346  max mem: 26157
Eval (hcp-train-subset): [58]  [61/62]  eta: 0:00:00  loss: 0.8173 (0.8186)  time: 0.1220  data: 0.0393  max mem: 26157
Eval (hcp-train-subset): [58] Total time: 0:00:12 (0.2078 s / it)
Averaged stats (hcp-train-subset): loss: 0.8173 (0.8186)
Making plots (hcp-train-subset): example=27
Eval (hcp-val): [58]  [ 0/62]  eta: 0:03:53  loss: 0.8219 (0.8219)  time: 3.7597  data: 3.6590  max mem: 26157
Eval (hcp-val): [58]  [61/62]  eta: 0:00:00  loss: 0.8233 (0.8247)  time: 0.1250  data: 0.0404  max mem: 26157
Eval (hcp-val): [58] Total time: 0:00:12 (0.2096 s / it)
Averaged stats (hcp-val): loss: 0.8233 (0.8247)
Making plots (hcp-val): example=61
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [59]  [   0/6250]  eta: 11:29:21  lr: 0.000049  grad: 0.1681 (0.1681)  loss: 0.8675 (0.8675)  time: 6.6178  data: 6.3411  max mem: 26157
Train: [59]  [ 100/6250]  eta: 0:34:08  lr: 0.000049  grad: 0.1502 (0.2012)  loss: 0.8137 (0.8111)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [59]  [ 200/6250]  eta: 0:30:26  lr: 0.000049  grad: 0.1354 (0.1764)  loss: 0.8088 (0.8079)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [59]  [ 300/6250]  eta: 0:28:58  lr: 0.000049  grad: 0.1486 (0.1728)  loss: 0.8012 (0.8059)  time: 0.2744  data: 0.0002  max mem: 26157
Train: [59]  [ 400/6250]  eta: 0:27:57  lr: 0.000049  grad: 0.1424 (0.1709)  loss: 0.7905 (0.8037)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [59]  [ 500/6250]  eta: 0:27:10  lr: 0.000049  grad: 0.1535 (0.1706)  loss: 0.7941 (0.8018)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [59]  [ 600/6250]  eta: 0:26:28  lr: 0.000049  grad: 0.1360 (0.1692)  loss: 0.8000 (0.8011)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [59]  [ 700/6250]  eta: 0:26:56  lr: 0.000049  grad: 0.1386 (0.1664)  loss: 0.8002 (0.8007)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [59]  [ 800/6250]  eta: 0:26:22  lr: 0.000049  grad: 0.1345 (0.1643)  loss: 0.7969 (0.8004)  time: 0.2743  data: 0.0002  max mem: 26157
Train: [59]  [ 900/6250]  eta: 0:26:14  lr: 0.000049  grad: 0.1349 (0.1627)  loss: 0.7946 (0.8000)  time: 0.3778  data: 0.1061  max mem: 26157
Train: [59]  [1000/6250]  eta: 0:25:32  lr: 0.000049  grad: 0.1298 (0.1614)  loss: 0.7903 (0.7996)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [59]  [1100/6250]  eta: 0:24:53  lr: 0.000049  grad: 0.1400 (0.1612)  loss: 0.7928 (0.7991)  time: 0.2733  data: 0.0003  max mem: 26157
Train: [59]  [1200/6250]  eta: 0:24:24  lr: 0.000049  grad: 0.1331 (0.1609)  loss: 0.7935 (0.7987)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [59]  [1300/6250]  eta: 0:23:47  lr: 0.000049  grad: 0.1497 (0.1624)  loss: 0.7926 (0.7982)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [59]  [1400/6250]  eta: 0:23:12  lr: 0.000049  grad: 0.1467 (0.1632)  loss: 0.7915 (0.7978)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [59]  [1500/6250]  eta: 0:22:38  lr: 0.000049  grad: 0.1421 (0.1633)  loss: 0.7859 (0.7972)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [59]  [1600/6250]  eta: 0:22:05  lr: 0.000049  grad: 0.1339 (0.1627)  loss: 0.7938 (0.7968)  time: 0.2739  data: 0.0002  max mem: 26157
Train: [59]  [1700/6250]  eta: 0:21:58  lr: 0.000049  grad: 0.1344 (0.1630)  loss: 0.7897 (0.7963)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [59]  [1800/6250]  eta: 0:21:24  lr: 0.000049  grad: 0.1382 (0.1630)  loss: 0.7852 (0.7956)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [59]  [1900/6250]  eta: 0:20:50  lr: 0.000049  grad: 0.1412 (0.1630)  loss: 0.7867 (0.7951)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [59]  [2000/6250]  eta: 0:20:17  lr: 0.000049  grad: 0.1469 (0.1633)  loss: 0.7817 (0.7947)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [59]  [2100/6250]  eta: 0:19:46  lr: 0.000048  grad: 0.1326 (0.1633)  loss: 0.7937 (0.7943)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [59]  [2200/6250]  eta: 0:19:14  lr: 0.000048  grad: 0.1519 (0.1636)  loss: 0.7889 (0.7941)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [59]  [2300/6250]  eta: 0:18:43  lr: 0.000048  grad: 0.1463 (0.1650)  loss: 0.7854 (0.7939)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [59]  [2400/6250]  eta: 0:18:12  lr: 0.000048  grad: 0.1594 (0.1647)  loss: 0.7930 (0.7936)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [59]  [2500/6250]  eta: 0:17:42  lr: 0.000048  grad: 0.1398 (0.1647)  loss: 0.7819 (0.7932)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [59]  [2600/6250]  eta: 0:17:12  lr: 0.000048  grad: 0.1441 (0.1645)  loss: 0.7807 (0.7930)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [59]  [2700/6250]  eta: 0:16:42  lr: 0.000048  grad: 0.1436 (0.1648)  loss: 0.7845 (0.7927)  time: 0.2719  data: 0.0003  max mem: 26157
Train: [59]  [2800/6250]  eta: 0:16:12  lr: 0.000048  grad: 0.1468 (0.1646)  loss: 0.7928 (0.7924)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [59]  [2900/6250]  eta: 0:15:46  lr: 0.000048  grad: 0.1404 (0.1647)  loss: 0.7903 (0.7923)  time: 0.4123  data: 0.1355  max mem: 26157
Train: [59]  [3000/6250]  eta: 0:15:27  lr: 0.000048  grad: 0.1562 (0.1649)  loss: 0.7872 (0.7921)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [59]  [3100/6250]  eta: 0:14:56  lr: 0.000048  grad: 0.1477 (0.1651)  loss: 0.7872 (0.7920)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [59]  [3200/6250]  eta: 0:14:27  lr: 0.000048  grad: 0.1477 (0.1650)  loss: 0.7958 (0.7919)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [59]  [3300/6250]  eta: 0:13:57  lr: 0.000048  grad: 0.1454 (0.1656)  loss: 0.7860 (0.7918)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [59]  [3400/6250]  eta: 0:13:27  lr: 0.000048  grad: 0.1475 (0.1659)  loss: 0.7948 (0.7917)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [59]  [3500/6250]  eta: 0:12:58  lr: 0.000048  grad: 0.1500 (0.1664)  loss: 0.7838 (0.7915)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [59]  [3600/6250]  eta: 0:12:29  lr: 0.000048  grad: 0.1499 (0.1664)  loss: 0.7834 (0.7915)  time: 0.2737  data: 0.0002  max mem: 26157
Train: [59]  [3700/6250]  eta: 0:11:59  lr: 0.000048  grad: 0.1560 (0.1670)  loss: 0.7865 (0.7914)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [59]  [3800/6250]  eta: 0:11:30  lr: 0.000048  grad: 0.1488 (0.1678)  loss: 0.7919 (0.7913)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [59]  [3900/6250]  eta: 0:11:10  lr: 0.000048  grad: 0.1494 (0.1680)  loss: 0.7861 (0.7913)  time: 0.2915  data: 0.0003  max mem: 26157
Train: [59]  [4000/6250]  eta: 0:10:41  lr: 0.000048  grad: 0.1533 (0.1684)  loss: 0.7936 (0.7911)  time: 0.2726  data: 0.0002  max mem: 26157
Train: [59]  [4100/6250]  eta: 0:10:12  lr: 0.000048  grad: 0.1584 (0.1690)  loss: 0.7837 (0.7910)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [59]  [4200/6250]  eta: 0:09:43  lr: 0.000048  grad: 0.1492 (0.1689)  loss: 0.7780 (0.7908)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [59]  [4300/6250]  eta: 0:09:14  lr: 0.000048  grad: 0.1476 (0.1692)  loss: 0.7847 (0.7906)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [59]  [4400/6250]  eta: 0:08:45  lr: 0.000048  grad: 0.1510 (0.1700)  loss: 0.7832 (0.7905)  time: 0.2765  data: 0.0002  max mem: 26157
Train: [59]  [4500/6250]  eta: 0:08:17  lr: 0.000048  grad: 0.1398 (0.1703)  loss: 0.7891 (0.7904)  time: 0.3919  data: 0.1226  max mem: 26157
Train: [59]  [4600/6250]  eta: 0:07:53  lr: 0.000048  grad: 0.1384 (0.1705)  loss: 0.7875 (0.7903)  time: 0.5807  data: 0.3038  max mem: 26157
Train: [59]  [4700/6250]  eta: 0:07:23  lr: 0.000048  grad: 0.1377 (0.1708)  loss: 0.7880 (0.7902)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [59]  [4800/6250]  eta: 0:06:56  lr: 0.000048  grad: 0.1536 (0.1715)  loss: 0.7879 (0.7902)  time: 0.4791  data: 0.2085  max mem: 26157
Train: [59]  [4900/6250]  eta: 0:06:27  lr: 0.000048  grad: 0.1405 (0.1713)  loss: 0.7919 (0.7902)  time: 0.2746  data: 0.0002  max mem: 26157
Train: [59]  [5000/6250]  eta: 0:05:58  lr: 0.000048  grad: 0.1495 (0.1716)  loss: 0.7889 (0.7901)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [59]  [5100/6250]  eta: 0:05:31  lr: 0.000048  grad: 0.1496 (0.1724)  loss: 0.7878 (0.7901)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [59]  [5200/6250]  eta: 0:05:02  lr: 0.000047  grad: 0.1473 (0.1724)  loss: 0.7910 (0.7901)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [59]  [5300/6250]  eta: 0:04:33  lr: 0.000047  grad: 0.1407 (0.1722)  loss: 0.7948 (0.7901)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [59]  [5400/6250]  eta: 0:04:04  lr: 0.000047  grad: 0.1472 (0.1726)  loss: 0.7842 (0.7901)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [59]  [5500/6250]  eta: 0:03:35  lr: 0.000047  grad: 0.1486 (0.1729)  loss: 0.7807 (0.7901)  time: 0.2697  data: 0.0001  max mem: 26157
Train: [59]  [5600/6250]  eta: 0:03:08  lr: 0.000047  grad: 0.1412 (0.1734)  loss: 0.7851 (0.7901)  time: 1.2379  data: 0.9620  max mem: 26157
Train: [59]  [5700/6250]  eta: 0:02:39  lr: 0.000047  grad: 0.1509 (0.1732)  loss: 0.7910 (0.7901)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [59]  [5800/6250]  eta: 0:02:10  lr: 0.000047  grad: 0.1430 (0.1733)  loss: 0.7907 (0.7901)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [59]  [5900/6250]  eta: 0:01:41  lr: 0.000047  grad: 0.1404 (0.1733)  loss: 0.7894 (0.7901)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [59]  [6000/6250]  eta: 0:01:12  lr: 0.000047  grad: 0.1518 (0.1732)  loss: 0.7906 (0.7901)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [59]  [6100/6250]  eta: 0:00:43  lr: 0.000047  grad: 0.1438 (0.1733)  loss: 0.7970 (0.7901)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [59]  [6200/6250]  eta: 0:00:14  lr: 0.000047  grad: 0.1570 (0.1735)  loss: 0.7866 (0.7901)  time: 0.2761  data: 0.0002  max mem: 26157
Train: [59]  [6249/6250]  eta: 0:00:00  lr: 0.000047  grad: 0.1541 (0.1736)  loss: 0.7885 (0.7901)  time: 0.2716  data: 0.0002  max mem: 26157
Train: [59] Total time: 0:30:07 (0.2893 s / it)
Averaged stats: lr: 0.000047  grad: 0.1541 (0.1736)  loss: 0.7885 (0.7901)
Eval (hcp-train-subset): [59]  [ 0/62]  eta: 0:04:14  loss: 0.8277 (0.8277)  time: 4.1119  data: 4.0094  max mem: 26157
Eval (hcp-train-subset): [59]  [61/62]  eta: 0:00:00  loss: 0.8145 (0.8174)  time: 0.1264  data: 0.0416  max mem: 26157
Eval (hcp-train-subset): [59] Total time: 0:00:13 (0.2195 s / it)
Averaged stats (hcp-train-subset): loss: 0.8145 (0.8174)
Making plots (hcp-train-subset): example=60
Eval (hcp-val): [59]  [ 0/62]  eta: 0:05:52  loss: 0.8198 (0.8198)  time: 5.6776  data: 5.5941  max mem: 26157
Eval (hcp-val): [59]  [61/62]  eta: 0:00:00  loss: 0.8244 (0.8252)  time: 0.1366  data: 0.0538  max mem: 26157
Eval (hcp-val): [59] Total time: 0:00:13 (0.2187 s / it)
Averaged stats (hcp-val): loss: 0.8244 (0.8252)
Making plots (hcp-val): example=14
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [60]  [   0/6250]  eta: 10:44:20  lr: 0.000047  grad: 0.1166 (0.1166)  loss: 0.8679 (0.8679)  time: 6.1857  data: 5.9121  max mem: 26157
Train: [60]  [ 100/6250]  eta: 0:33:50  lr: 0.000047  grad: 0.1402 (0.1793)  loss: 0.8028 (0.8089)  time: 0.2725  data: 0.0002  max mem: 26157
Train: [60]  [ 200/6250]  eta: 0:30:22  lr: 0.000047  grad: 0.1476 (0.1717)  loss: 0.7960 (0.8067)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [60]  [ 300/6250]  eta: 0:28:52  lr: 0.000047  grad: 0.1530 (0.1712)  loss: 0.7945 (0.8039)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [60]  [ 400/6250]  eta: 0:27:52  lr: 0.000047  grad: 0.1593 (0.1737)  loss: 0.7982 (0.8018)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [60]  [ 500/6250]  eta: 0:27:08  lr: 0.000047  grad: 0.1524 (0.1761)  loss: 0.7979 (0.8002)  time: 0.2757  data: 0.0004  max mem: 26157
Train: [60]  [ 600/6250]  eta: 0:27:03  lr: 0.000047  grad: 0.1488 (0.1763)  loss: 0.7880 (0.7991)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [60]  [ 700/6250]  eta: 0:26:21  lr: 0.000047  grad: 0.1492 (0.1770)  loss: 0.7936 (0.7986)  time: 0.2711  data: 0.0005  max mem: 26157
Train: [60]  [ 800/6250]  eta: 0:25:42  lr: 0.000047  grad: 0.1457 (0.1766)  loss: 0.7863 (0.7981)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [60]  [ 900/6250]  eta: 0:25:05  lr: 0.000047  grad: 0.1476 (0.1785)  loss: 0.8021 (0.7976)  time: 0.2699  data: 0.0003  max mem: 26157
Train: [60]  [1000/6250]  eta: 0:24:31  lr: 0.000047  grad: 0.1366 (0.1765)  loss: 0.7929 (0.7973)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [60]  [1100/6250]  eta: 0:23:58  lr: 0.000047  grad: 0.1657 (0.1754)  loss: 0.7896 (0.7967)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [60]  [1200/6250]  eta: 0:23:26  lr: 0.000047  grad: 0.1407 (0.1750)  loss: 0.7832 (0.7962)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [60]  [1300/6250]  eta: 0:22:55  lr: 0.000047  grad: 0.1530 (0.1745)  loss: 0.7873 (0.7959)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [60]  [1400/6250]  eta: 0:22:25  lr: 0.000047  grad: 0.1475 (0.1741)  loss: 0.7844 (0.7956)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [60]  [1500/6250]  eta: 0:21:55  lr: 0.000047  grad: 0.1431 (0.1740)  loss: 0.7982 (0.7954)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [60]  [1600/6250]  eta: 0:21:25  lr: 0.000047  grad: 0.1380 (0.1739)  loss: 0.7899 (0.7950)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [60]  [1700/6250]  eta: 0:20:56  lr: 0.000047  grad: 0.1377 (0.1730)  loss: 0.7963 (0.7949)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [60]  [1800/6250]  eta: 0:20:27  lr: 0.000047  grad: 0.1393 (0.1724)  loss: 0.7818 (0.7946)  time: 0.2759  data: 0.0003  max mem: 26157
Train: [60]  [1900/6250]  eta: 0:19:59  lr: 0.000047  grad: 0.1394 (0.1725)  loss: 0.7950 (0.7945)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [60]  [2000/6250]  eta: 0:19:31  lr: 0.000047  grad: 0.1474 (0.1735)  loss: 0.7923 (0.7943)  time: 0.2783  data: 0.0002  max mem: 26157
Train: [60]  [2100/6250]  eta: 0:19:02  lr: 0.000046  grad: 0.1578 (0.1730)  loss: 0.7938 (0.7942)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [60]  [2200/6250]  eta: 0:18:34  lr: 0.000046  grad: 0.1483 (0.1739)  loss: 0.7924 (0.7942)  time: 0.2803  data: 0.0002  max mem: 26157
Train: [60]  [2300/6250]  eta: 0:18:12  lr: 0.000046  grad: 0.1566 (0.1745)  loss: 0.7912 (0.7942)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [60]  [2400/6250]  eta: 0:17:58  lr: 0.000046  grad: 0.1673 (0.1756)  loss: 0.7902 (0.7940)  time: 0.5318  data: 0.2515  max mem: 26157
Train: [60]  [2500/6250]  eta: 0:17:29  lr: 0.000046  grad: 0.1356 (0.1753)  loss: 0.7890 (0.7939)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [60]  [2600/6250]  eta: 0:17:00  lr: 0.000046  grad: 0.1389 (0.1750)  loss: 0.7943 (0.7938)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [60]  [2700/6250]  eta: 0:16:30  lr: 0.000046  grad: 0.1452 (0.1754)  loss: 0.7983 (0.7938)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [60]  [2800/6250]  eta: 0:16:01  lr: 0.000046  grad: 0.1464 (0.1757)  loss: 0.7969 (0.7939)  time: 0.2692  data: 0.0001  max mem: 26157
Train: [60]  [2900/6250]  eta: 0:15:36  lr: 0.000046  grad: 0.1634 (0.1768)  loss: 0.7918 (0.7940)  time: 0.4052  data: 0.1321  max mem: 26157
Train: [60]  [3000/6250]  eta: 0:15:07  lr: 0.000046  grad: 0.1511 (0.1774)  loss: 0.7964 (0.7940)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [60]  [3100/6250]  eta: 0:14:38  lr: 0.000046  grad: 0.1469 (0.1778)  loss: 0.7902 (0.7940)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [60]  [3200/6250]  eta: 0:14:09  lr: 0.000046  grad: 0.1356 (0.1781)  loss: 0.7921 (0.7939)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [60]  [3300/6250]  eta: 0:13:40  lr: 0.000046  grad: 0.1521 (0.1777)  loss: 0.7918 (0.7940)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [60]  [3400/6250]  eta: 0:13:12  lr: 0.000046  grad: 0.1492 (0.1774)  loss: 0.7894 (0.7939)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [60]  [3500/6250]  eta: 0:12:43  lr: 0.000046  grad: 0.1427 (0.1768)  loss: 0.7963 (0.7940)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [60]  [3600/6250]  eta: 0:12:15  lr: 0.000046  grad: 0.1442 (0.1765)  loss: 0.7939 (0.7940)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [60]  [3700/6250]  eta: 0:11:47  lr: 0.000046  grad: 0.1500 (0.1764)  loss: 0.7907 (0.7939)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [60]  [3800/6250]  eta: 0:11:19  lr: 0.000046  grad: 0.1397 (0.1758)  loss: 0.7897 (0.7939)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [60]  [3900/6250]  eta: 0:10:50  lr: 0.000046  grad: 0.1378 (0.1758)  loss: 0.7997 (0.7940)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [60]  [4000/6250]  eta: 0:10:22  lr: 0.000046  grad: 0.1472 (0.1758)  loss: 0.7944 (0.7940)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [60]  [4100/6250]  eta: 0:09:54  lr: 0.000046  grad: 0.1577 (0.1762)  loss: 0.7937 (0.7939)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [60]  [4200/6250]  eta: 0:09:29  lr: 0.000046  grad: 0.1460 (0.1770)  loss: 0.7900 (0.7940)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [60]  [4300/6250]  eta: 0:09:01  lr: 0.000046  grad: 0.1411 (0.1769)  loss: 0.7946 (0.7939)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [60]  [4400/6250]  eta: 0:08:33  lr: 0.000046  grad: 0.1425 (0.1769)  loss: 0.7949 (0.7939)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [60]  [4500/6250]  eta: 0:08:05  lr: 0.000046  grad: 0.1516 (0.1769)  loss: 0.7879 (0.7938)  time: 0.3212  data: 0.0501  max mem: 26157
Train: [60]  [4600/6250]  eta: 0:07:37  lr: 0.000046  grad: 0.1441 (0.1769)  loss: 0.7883 (0.7938)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [60]  [4700/6250]  eta: 0:07:09  lr: 0.000046  grad: 0.1508 (0.1773)  loss: 0.7832 (0.7937)  time: 0.2733  data: 0.0002  max mem: 26157
Train: [60]  [4800/6250]  eta: 0:06:41  lr: 0.000046  grad: 0.1564 (0.1774)  loss: 0.7892 (0.7936)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [60]  [4900/6250]  eta: 0:06:14  lr: 0.000046  grad: 0.1781 (0.1798)  loss: 0.7896 (0.7935)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [60]  [5000/6250]  eta: 0:05:46  lr: 0.000046  grad: 0.1480 (0.1795)  loss: 0.7902 (0.7934)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [60]  [5100/6250]  eta: 0:05:18  lr: 0.000046  grad: 0.1507 (0.1797)  loss: 0.7917 (0.7933)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [60]  [5200/6250]  eta: 0:04:50  lr: 0.000045  grad: 0.1474 (0.1797)  loss: 0.7926 (0.7932)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [60]  [5300/6250]  eta: 0:04:22  lr: 0.000045  grad: 0.1557 (0.1800)  loss: 0.7913 (0.7931)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [60]  [5400/6250]  eta: 0:03:54  lr: 0.000045  grad: 0.1566 (0.1802)  loss: 0.7932 (0.7932)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [60]  [5500/6250]  eta: 0:03:27  lr: 0.000045  grad: 0.1469 (0.1802)  loss: 0.7930 (0.7932)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [60]  [5600/6250]  eta: 0:02:59  lr: 0.000045  grad: 0.1471 (0.1807)  loss: 0.7959 (0.7931)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [60]  [5700/6250]  eta: 0:02:31  lr: 0.000045  grad: 0.1593 (0.1814)  loss: 0.7876 (0.7931)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [60]  [5800/6250]  eta: 0:02:04  lr: 0.000045  grad: 0.1524 (0.1815)  loss: 0.7845 (0.7930)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [60]  [5900/6250]  eta: 0:01:36  lr: 0.000045  grad: 0.1499 (0.1817)  loss: 0.7855 (0.7930)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [60]  [6000/6250]  eta: 0:01:08  lr: 0.000045  grad: 0.1528 (0.1816)  loss: 0.7947 (0.7929)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [60]  [6100/6250]  eta: 0:00:41  lr: 0.000045  grad: 0.1540 (0.1814)  loss: 0.7898 (0.7929)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [60]  [6200/6250]  eta: 0:00:13  lr: 0.000045  grad: 0.1520 (0.1819)  loss: 0.7897 (0.7928)  time: 0.2750  data: 0.0002  max mem: 26157
Train: [60]  [6249/6250]  eta: 0:00:00  lr: 0.000045  grad: 0.1518 (0.1819)  loss: 0.7929 (0.7928)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [60] Total time: 0:28:49 (0.2767 s / it)
Averaged stats: lr: 0.000045  grad: 0.1518 (0.1819)  loss: 0.7929 (0.7928)
Eval (hcp-train-subset): [60]  [ 0/62]  eta: 0:03:52  loss: 0.8252 (0.8252)  time: 3.7458  data: 3.6363  max mem: 26157
Eval (hcp-train-subset): [60]  [61/62]  eta: 0:00:00  loss: 0.8151 (0.8165)  time: 0.1209  data: 0.0380  max mem: 26157
Eval (hcp-train-subset): [60] Total time: 0:00:13 (0.2136 s / it)
Averaged stats (hcp-train-subset): loss: 0.8151 (0.8165)
Making plots (hcp-train-subset): example=10
Eval (hcp-val): [60]  [ 0/62]  eta: 0:05:32  loss: 0.8185 (0.8185)  time: 5.3575  data: 5.2728  max mem: 26157
Eval (hcp-val): [60]  [61/62]  eta: 0:00:00  loss: 0.8254 (0.8248)  time: 0.1259  data: 0.0413  max mem: 26157
Eval (hcp-val): [60] Total time: 0:00:13 (0.2127 s / it)
Averaged stats (hcp-val): loss: 0.8254 (0.8248)
Making plots (hcp-val): example=57
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [61]  [   0/6250]  eta: 11:00:24  lr: 0.000045  grad: 0.1489 (0.1489)  loss: 0.8429 (0.8429)  time: 6.3399  data: 6.0610  max mem: 26157
Train: [61]  [ 100/6250]  eta: 0:34:17  lr: 0.000045  grad: 0.1611 (0.2136)  loss: 0.8087 (0.8137)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [61]  [ 200/6250]  eta: 0:30:30  lr: 0.000045  grad: 0.1849 (0.2110)  loss: 0.8004 (0.8049)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [61]  [ 300/6250]  eta: 0:28:55  lr: 0.000045  grad: 0.1431 (0.2046)  loss: 0.7950 (0.8022)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [61]  [ 400/6250]  eta: 0:27:53  lr: 0.000045  grad: 0.1646 (0.2021)  loss: 0.7888 (0.8012)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [61]  [ 500/6250]  eta: 0:27:05  lr: 0.000045  grad: 0.1489 (0.1997)  loss: 0.7958 (0.8004)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [61]  [ 600/6250]  eta: 0:26:27  lr: 0.000045  grad: 0.1542 (0.1968)  loss: 0.8049 (0.7999)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [61]  [ 700/6250]  eta: 0:25:52  lr: 0.000045  grad: 0.1727 (0.1966)  loss: 0.8039 (0.7995)  time: 0.2738  data: 0.0002  max mem: 26157
Train: [61]  [ 800/6250]  eta: 0:26:03  lr: 0.000045  grad: 0.1440 (0.1942)  loss: 0.7894 (0.7995)  time: 0.2733  data: 0.0004  max mem: 26157
Train: [61]  [ 900/6250]  eta: 0:25:59  lr: 0.000045  grad: 0.1470 (0.1934)  loss: 0.7988 (0.7992)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [61]  [1000/6250]  eta: 0:25:28  lr: 0.000045  grad: 0.1545 (0.1912)  loss: 0.7939 (0.7987)  time: 0.2735  data: 0.0002  max mem: 26157
Train: [61]  [1100/6250]  eta: 0:24:49  lr: 0.000045  grad: 0.1608 (0.1901)  loss: 0.7924 (0.7981)  time: 0.2716  data: 0.0003  max mem: 26157
Train: [61]  [1200/6250]  eta: 0:24:30  lr: 0.000045  grad: 0.1586 (0.1890)  loss: 0.7917 (0.7975)  time: 0.3384  data: 0.0675  max mem: 26157
Train: [61]  [1300/6250]  eta: 0:23:53  lr: 0.000045  grad: 0.1503 (0.1871)  loss: 0.7933 (0.7970)  time: 0.2759  data: 0.0003  max mem: 26157
Train: [61]  [1400/6250]  eta: 0:23:25  lr: 0.000045  grad: 0.1461 (0.1858)  loss: 0.7892 (0.7965)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [61]  [1500/6250]  eta: 0:22:50  lr: 0.000045  grad: 0.1524 (0.1867)  loss: 0.7888 (0.7960)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [61]  [1600/6250]  eta: 0:22:15  lr: 0.000045  grad: 0.1594 (0.1862)  loss: 0.7930 (0.7956)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [61]  [1700/6250]  eta: 0:21:42  lr: 0.000045  grad: 0.1587 (0.1879)  loss: 0.7848 (0.7952)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [61]  [1800/6250]  eta: 0:21:09  lr: 0.000045  grad: 0.1507 (0.1886)  loss: 0.7922 (0.7949)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [61]  [1900/6250]  eta: 0:20:47  lr: 0.000045  grad: 0.1595 (0.1891)  loss: 0.7871 (0.7946)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [61]  [2000/6250]  eta: 0:20:17  lr: 0.000045  grad: 0.1732 (0.1894)  loss: 0.7851 (0.7945)  time: 0.3221  data: 0.0485  max mem: 26157
Train: [61]  [2100/6250]  eta: 0:19:45  lr: 0.000044  grad: 0.1500 (0.1906)  loss: 0.7899 (0.7942)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [61]  [2200/6250]  eta: 0:19:14  lr: 0.000044  grad: 0.1477 (0.1900)  loss: 0.7862 (0.7939)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [61]  [2300/6250]  eta: 0:18:43  lr: 0.000044  grad: 0.1463 (0.1890)  loss: 0.7898 (0.7938)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [61]  [2400/6250]  eta: 0:18:12  lr: 0.000044  grad: 0.1632 (0.1893)  loss: 0.7896 (0.7936)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [61]  [2500/6250]  eta: 0:17:41  lr: 0.000044  grad: 0.1503 (0.1898)  loss: 0.7932 (0.7935)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [61]  [2600/6250]  eta: 0:17:11  lr: 0.000044  grad: 0.1471 (0.1896)  loss: 0.7921 (0.7934)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [61]  [2700/6250]  eta: 0:16:41  lr: 0.000044  grad: 0.1830 (0.1897)  loss: 0.7860 (0.7932)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [61]  [2800/6250]  eta: 0:16:11  lr: 0.000044  grad: 0.1692 (0.1894)  loss: 0.7889 (0.7932)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [61]  [2900/6250]  eta: 0:15:42  lr: 0.000044  grad: 0.1364 (0.1890)  loss: 0.7931 (0.7931)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [61]  [3000/6250]  eta: 0:15:12  lr: 0.000044  grad: 0.1681 (0.1888)  loss: 0.7923 (0.7931)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [61]  [3100/6250]  eta: 0:14:43  lr: 0.000044  grad: 0.1493 (0.1888)  loss: 0.7845 (0.7930)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [61]  [3200/6250]  eta: 0:14:14  lr: 0.000044  grad: 0.1613 (0.1897)  loss: 0.7892 (0.7928)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [61]  [3300/6250]  eta: 0:13:45  lr: 0.000044  grad: 0.1788 (0.1902)  loss: 0.7849 (0.7928)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [61]  [3400/6250]  eta: 0:13:16  lr: 0.000044  grad: 0.2301 (0.1920)  loss: 0.7853 (0.7927)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [61]  [3500/6250]  eta: 0:12:48  lr: 0.000044  grad: 0.1485 (0.1919)  loss: 0.7929 (0.7927)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [61]  [3600/6250]  eta: 0:12:19  lr: 0.000044  grad: 0.1678 (0.1926)  loss: 0.7922 (0.7927)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [61]  [3700/6250]  eta: 0:11:51  lr: 0.000044  grad: 0.1651 (0.1929)  loss: 0.7882 (0.7926)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [61]  [3800/6250]  eta: 0:11:22  lr: 0.000044  grad: 0.1566 (0.1929)  loss: 0.7857 (0.7926)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [61]  [3900/6250]  eta: 0:10:54  lr: 0.000044  grad: 0.1927 (0.1933)  loss: 0.7862 (0.7926)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [61]  [4000/6250]  eta: 0:10:26  lr: 0.000044  grad: 0.1587 (0.1935)  loss: 0.7899 (0.7926)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [61]  [4100/6250]  eta: 0:09:57  lr: 0.000044  grad: 0.1575 (0.1935)  loss: 0.7968 (0.7926)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [61]  [4200/6250]  eta: 0:09:29  lr: 0.000044  grad: 0.1744 (0.1938)  loss: 0.7918 (0.7926)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [61]  [4300/6250]  eta: 0:09:01  lr: 0.000044  grad: 0.2041 (0.1943)  loss: 0.7956 (0.7926)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [61]  [4400/6250]  eta: 0:08:33  lr: 0.000044  grad: 0.1635 (0.1950)  loss: 0.7885 (0.7926)  time: 0.2723  data: 0.0002  max mem: 26157
Train: [61]  [4500/6250]  eta: 0:08:05  lr: 0.000044  grad: 0.1580 (0.1947)  loss: 0.7905 (0.7925)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [61]  [4600/6250]  eta: 0:07:37  lr: 0.000044  grad: 0.1618 (0.1944)  loss: 0.7863 (0.7925)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [61]  [4700/6250]  eta: 0:07:09  lr: 0.000044  grad: 0.1613 (0.1943)  loss: 0.7920 (0.7925)  time: 0.3148  data: 0.0359  max mem: 26157
Train: [61]  [4800/6250]  eta: 0:06:41  lr: 0.000044  grad: 0.1437 (0.1942)  loss: 0.7939 (0.7925)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [61]  [4900/6250]  eta: 0:06:13  lr: 0.000044  grad: 0.1541 (0.1940)  loss: 0.7915 (0.7925)  time: 0.2714  data: 0.0002  max mem: 26157
Train: [61]  [5000/6250]  eta: 0:05:45  lr: 0.000044  grad: 0.1535 (0.1936)  loss: 0.7936 (0.7924)  time: 0.2740  data: 0.0002  max mem: 26157
Train: [61]  [5100/6250]  eta: 0:05:18  lr: 0.000044  grad: 0.1667 (0.1934)  loss: 0.7883 (0.7924)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [61]  [5200/6250]  eta: 0:04:50  lr: 0.000044  grad: 0.1673 (0.1930)  loss: 0.7870 (0.7924)  time: 0.2750  data: 0.0002  max mem: 26157
Train: [61]  [5300/6250]  eta: 0:04:22  lr: 0.000043  grad: 0.1724 (0.1934)  loss: 0.7934 (0.7924)  time: 0.2753  data: 0.0002  max mem: 26157
Train: [61]  [5400/6250]  eta: 0:03:54  lr: 0.000043  grad: 0.1690 (0.1938)  loss: 0.7954 (0.7924)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [61]  [5500/6250]  eta: 0:03:27  lr: 0.000043  grad: 0.1548 (0.1939)  loss: 0.7944 (0.7923)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [61]  [5600/6250]  eta: 0:03:00  lr: 0.000043  grad: 0.1628 (0.1939)  loss: 0.7971 (0.7923)  time: 0.5896  data: 0.3198  max mem: 26157
Train: [61]  [5700/6250]  eta: 0:02:32  lr: 0.000043  grad: 0.1583 (0.1942)  loss: 0.7812 (0.7923)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [61]  [5800/6250]  eta: 0:02:05  lr: 0.000043  grad: 0.1798 (0.1942)  loss: 0.7854 (0.7922)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [61]  [5900/6250]  eta: 0:01:37  lr: 0.000043  grad: 0.1556 (0.1942)  loss: 0.7887 (0.7922)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [61]  [6000/6250]  eta: 0:01:09  lr: 0.000043  grad: 0.1527 (0.1944)  loss: 0.7916 (0.7922)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [61]  [6100/6250]  eta: 0:00:41  lr: 0.000043  grad: 0.1588 (0.1944)  loss: 0.7868 (0.7921)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [61]  [6200/6250]  eta: 0:00:13  lr: 0.000043  grad: 0.1507 (0.1944)  loss: 0.7893 (0.7921)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [61]  [6249/6250]  eta: 0:00:00  lr: 0.000043  grad: 0.1450 (0.1945)  loss: 0.7887 (0.7921)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [61] Total time: 0:29:01 (0.2786 s / it)
Averaged stats: lr: 0.000043  grad: 0.1450 (0.1945)  loss: 0.7887 (0.7921)
Eval (hcp-train-subset): [61]  [ 0/62]  eta: 0:04:53  loss: 0.8256 (0.8256)  time: 4.7362  data: 4.6522  max mem: 26157
Eval (hcp-train-subset): [61]  [61/62]  eta: 0:00:00  loss: 0.8149 (0.8169)  time: 0.1235  data: 0.0405  max mem: 26157
Eval (hcp-train-subset): [61] Total time: 0:00:13 (0.2104 s / it)
Averaged stats (hcp-train-subset): loss: 0.8149 (0.8169)
Making plots (hcp-train-subset): example=36
Eval (hcp-val): [61]  [ 0/62]  eta: 0:04:23  loss: 0.8239 (0.8239)  time: 4.2540  data: 4.1454  max mem: 26157
Eval (hcp-val): [61]  [61/62]  eta: 0:00:00  loss: 0.8239 (0.8254)  time: 0.1342  data: 0.0513  max mem: 26157
Eval (hcp-val): [61] Total time: 0:00:13 (0.2193 s / it)
Averaged stats (hcp-val): loss: 0.8239 (0.8254)
Making plots (hcp-val): example=27
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [62]  [   0/6250]  eta: 11:49:07  lr: 0.000043  grad: 0.7045 (0.7045)  loss: 0.7668 (0.7668)  time: 6.8076  data: 6.5280  max mem: 26157
Train: [62]  [ 100/6250]  eta: 0:34:31  lr: 0.000043  grad: 0.1666 (0.2059)  loss: 0.7945 (0.8088)  time: 0.2719  data: 0.0002  max mem: 26157
Train: [62]  [ 200/6250]  eta: 0:30:45  lr: 0.000043  grad: 0.1904 (0.1958)  loss: 0.7957 (0.8039)  time: 0.2721  data: 0.0002  max mem: 26157
Train: [62]  [ 300/6250]  eta: 0:29:07  lr: 0.000043  grad: 0.1766 (0.1936)  loss: 0.7935 (0.7996)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [62]  [ 400/6250]  eta: 0:28:05  lr: 0.000043  grad: 0.1524 (0.1931)  loss: 0.7928 (0.7966)  time: 0.2728  data: 0.0002  max mem: 26157
Train: [62]  [ 500/6250]  eta: 0:27:15  lr: 0.000043  grad: 0.1523 (0.1961)  loss: 0.7860 (0.7945)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [62]  [ 600/6250]  eta: 0:26:33  lr: 0.000043  grad: 0.1529 (0.1964)  loss: 0.7880 (0.7937)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [62]  [ 700/6250]  eta: 0:25:55  lr: 0.000043  grad: 0.1532 (0.1960)  loss: 0.7944 (0.7935)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [62]  [ 800/6250]  eta: 0:25:19  lr: 0.000043  grad: 0.1871 (0.1977)  loss: 0.7949 (0.7930)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [62]  [ 900/6250]  eta: 0:24:46  lr: 0.000043  grad: 0.1463 (0.1956)  loss: 0.7935 (0.7928)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [62]  [1000/6250]  eta: 0:24:14  lr: 0.000043  grad: 0.1704 (0.1967)  loss: 0.8035 (0.7928)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [62]  [1100/6250]  eta: 0:23:43  lr: 0.000043  grad: 0.1588 (0.1982)  loss: 0.7879 (0.7925)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [62]  [1200/6250]  eta: 0:23:12  lr: 0.000043  grad: 0.1624 (0.1987)  loss: 0.7984 (0.7921)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [62]  [1300/6250]  eta: 0:22:41  lr: 0.000043  grad: 0.1526 (0.1966)  loss: 0.7906 (0.7918)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [62]  [1400/6250]  eta: 0:22:12  lr: 0.000043  grad: 0.1527 (0.1956)  loss: 0.7914 (0.7917)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [62]  [1500/6250]  eta: 0:21:42  lr: 0.000043  grad: 0.1478 (0.1966)  loss: 0.7883 (0.7914)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [62]  [1600/6250]  eta: 0:21:13  lr: 0.000043  grad: 0.1621 (0.1969)  loss: 0.7967 (0.7913)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [62]  [1700/6250]  eta: 0:20:44  lr: 0.000043  grad: 0.1589 (0.1961)  loss: 0.7874 (0.7910)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [62]  [1800/6250]  eta: 0:20:15  lr: 0.000043  grad: 0.1737 (0.1971)  loss: 0.7911 (0.7909)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [62]  [1900/6250]  eta: 0:19:47  lr: 0.000043  grad: 0.1520 (0.1969)  loss: 0.7882 (0.7908)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [62]  [2000/6250]  eta: 0:19:19  lr: 0.000043  grad: 0.1614 (0.1969)  loss: 0.7915 (0.7907)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [62]  [2100/6250]  eta: 0:18:51  lr: 0.000043  grad: 0.1542 (0.1969)  loss: 0.7834 (0.7906)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [62]  [2200/6250]  eta: 0:18:23  lr: 0.000042  grad: 0.1680 (0.1971)  loss: 0.7825 (0.7904)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [62]  [2300/6250]  eta: 0:17:55  lr: 0.000042  grad: 0.1529 (0.1965)  loss: 0.7829 (0.7901)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [62]  [2400/6250]  eta: 0:17:27  lr: 0.000042  grad: 0.1548 (0.1960)  loss: 0.7890 (0.7899)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [62]  [2500/6250]  eta: 0:16:59  lr: 0.000042  grad: 0.1547 (0.1959)  loss: 0.7886 (0.7897)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [62]  [2600/6250]  eta: 0:16:32  lr: 0.000042  grad: 0.1508 (0.1961)  loss: 0.7858 (0.7896)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [62]  [2700/6250]  eta: 0:16:04  lr: 0.000042  grad: 0.1566 (0.1963)  loss: 0.7899 (0.7895)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [62]  [2800/6250]  eta: 0:15:36  lr: 0.000042  grad: 0.1564 (0.1973)  loss: 0.7838 (0.7893)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [62]  [2900/6250]  eta: 0:15:09  lr: 0.000042  grad: 0.1584 (0.1969)  loss: 0.7728 (0.7891)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [62]  [3000/6250]  eta: 0:14:41  lr: 0.000042  grad: 0.1651 (0.1971)  loss: 0.7884 (0.7890)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [62]  [3100/6250]  eta: 0:14:14  lr: 0.000042  grad: 0.1422 (0.1968)  loss: 0.7939 (0.7890)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [62]  [3200/6250]  eta: 0:13:47  lr: 0.000042  grad: 0.1814 (0.1970)  loss: 0.7826 (0.7890)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [62]  [3300/6250]  eta: 0:13:19  lr: 0.000042  grad: 0.1626 (0.1972)  loss: 0.7846 (0.7889)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [62]  [3400/6250]  eta: 0:12:52  lr: 0.000042  grad: 0.1575 (0.1971)  loss: 0.7904 (0.7888)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [62]  [3500/6250]  eta: 0:12:25  lr: 0.000042  grad: 0.1587 (0.1973)  loss: 0.7904 (0.7888)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [62]  [3600/6250]  eta: 0:11:57  lr: 0.000042  grad: 0.1520 (0.1969)  loss: 0.7926 (0.7887)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [62]  [3700/6250]  eta: 0:11:30  lr: 0.000042  grad: 0.1595 (0.1975)  loss: 0.7861 (0.7886)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [62]  [3800/6250]  eta: 0:11:03  lr: 0.000042  grad: 0.1492 (0.1982)  loss: 0.7893 (0.7886)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [62]  [3900/6250]  eta: 0:10:36  lr: 0.000042  grad: 0.1455 (0.1972)  loss: 0.7885 (0.7885)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [62]  [4000/6250]  eta: 0:10:08  lr: 0.000042  grad: 0.1574 (0.1966)  loss: 0.7861 (0.7885)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [62]  [4100/6250]  eta: 0:09:41  lr: 0.000042  grad: 0.1596 (0.1973)  loss: 0.7869 (0.7884)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [62]  [4200/6250]  eta: 0:09:14  lr: 0.000042  grad: 0.1506 (0.1974)  loss: 0.7818 (0.7885)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [62]  [4300/6250]  eta: 0:08:47  lr: 0.000042  grad: 0.1530 (0.1977)  loss: 0.7883 (0.7885)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [62]  [4400/6250]  eta: 0:08:20  lr: 0.000042  grad: 0.1516 (0.1981)  loss: 0.7916 (0.7886)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [62]  [4500/6250]  eta: 0:07:53  lr: 0.000042  grad: 0.1472 (0.1983)  loss: 0.7905 (0.7886)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [62]  [4600/6250]  eta: 0:07:26  lr: 0.000042  grad: 0.1693 (0.1985)  loss: 0.7931 (0.7887)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [62]  [4700/6250]  eta: 0:06:59  lr: 0.000042  grad: 0.1652 (0.1992)  loss: 0.7896 (0.7888)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [62]  [4800/6250]  eta: 0:06:32  lr: 0.000042  grad: 0.1798 (0.1995)  loss: 0.7935 (0.7889)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [62]  [4900/6250]  eta: 0:06:05  lr: 0.000042  grad: 0.1646 (0.2005)  loss: 0.7982 (0.7890)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [62]  [5000/6250]  eta: 0:05:38  lr: 0.000042  grad: 0.1658 (0.2008)  loss: 0.7859 (0.7890)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [62]  [5100/6250]  eta: 0:05:11  lr: 0.000042  grad: 0.1725 (0.2014)  loss: 0.7867 (0.7889)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [62]  [5200/6250]  eta: 0:04:43  lr: 0.000042  grad: 0.1569 (0.2018)  loss: 0.7932 (0.7890)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [62]  [5300/6250]  eta: 0:04:16  lr: 0.000042  grad: 0.1772 (0.2024)  loss: 0.7858 (0.7890)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [62]  [5400/6250]  eta: 0:03:49  lr: 0.000041  grad: 0.1545 (0.2021)  loss: 0.7971 (0.7890)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [62]  [5500/6250]  eta: 0:03:22  lr: 0.000041  grad: 0.1551 (0.2021)  loss: 0.7877 (0.7889)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [62]  [5600/6250]  eta: 0:02:55  lr: 0.000041  grad: 0.1644 (0.2023)  loss: 0.7923 (0.7889)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [62]  [5700/6250]  eta: 0:02:28  lr: 0.000041  grad: 0.1524 (0.2020)  loss: 0.7880 (0.7889)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [62]  [5800/6250]  eta: 0:02:01  lr: 0.000041  grad: 0.1738 (0.2020)  loss: 0.7895 (0.7889)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [62]  [5900/6250]  eta: 0:01:34  lr: 0.000041  grad: 0.1671 (0.2022)  loss: 0.7920 (0.7889)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [62]  [6000/6250]  eta: 0:01:07  lr: 0.000041  grad: 0.1740 (0.2031)  loss: 0.7992 (0.7889)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [62]  [6100/6250]  eta: 0:00:40  lr: 0.000041  grad: 0.1611 (0.2030)  loss: 0.7877 (0.7890)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [62]  [6200/6250]  eta: 0:00:13  lr: 0.000041  grad: 0.1570 (0.2030)  loss: 0.7891 (0.7890)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [62]  [6249/6250]  eta: 0:00:00  lr: 0.000041  grad: 0.1587 (0.2030)  loss: 0.7887 (0.7890)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [62] Total time: 0:28:14 (0.2712 s / it)
Averaged stats: lr: 0.000041  grad: 0.1587 (0.2030)  loss: 0.7887 (0.7890)
Eval (hcp-train-subset): [62]  [ 0/62]  eta: 0:03:29  loss: 0.8228 (0.8228)  time: 3.3726  data: 3.2615  max mem: 26157
Eval (hcp-train-subset): [62]  [61/62]  eta: 0:00:00  loss: 0.8174 (0.8165)  time: 0.1021  data: 0.0174  max mem: 26157
Eval (hcp-train-subset): [62] Total time: 0:00:12 (0.1997 s / it)
Averaged stats (hcp-train-subset): loss: 0.8174 (0.8165)
Making plots (hcp-train-subset): example=45
Eval (hcp-val): [62]  [ 0/62]  eta: 0:04:45  loss: 0.8242 (0.8242)  time: 4.6059  data: 4.5214  max mem: 26157
Eval (hcp-val): [62]  [61/62]  eta: 0:00:00  loss: 0.8235 (0.8249)  time: 0.1298  data: 0.0452  max mem: 26157
Eval (hcp-val): [62] Total time: 0:00:13 (0.2106 s / it)
Averaged stats (hcp-val): loss: 0.8235 (0.8249)
Making plots (hcp-val): example=58
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [63]  [   0/6250]  eta: 9:08:42  lr: 0.000041  grad: 0.1585 (0.1585)  loss: 0.8331 (0.8331)  time: 5.2676  data: 4.9333  max mem: 26157
Train: [63]  [ 100/6250]  eta: 0:33:16  lr: 0.000041  grad: 0.1830 (0.2186)  loss: 0.7935 (0.7999)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [63]  [ 200/6250]  eta: 0:29:59  lr: 0.000041  grad: 0.1578 (0.2031)  loss: 0.7860 (0.7965)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [63]  [ 300/6250]  eta: 0:28:34  lr: 0.000041  grad: 0.1557 (0.1943)  loss: 0.7879 (0.7960)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [63]  [ 400/6250]  eta: 0:27:37  lr: 0.000041  grad: 0.1517 (0.1853)  loss: 0.7971 (0.7966)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [63]  [ 500/6250]  eta: 0:26:52  lr: 0.000041  grad: 0.1511 (0.1829)  loss: 0.7954 (0.7971)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [63]  [ 600/6250]  eta: 0:26:13  lr: 0.000041  grad: 0.1333 (0.1839)  loss: 0.8041 (0.7976)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [63]  [ 700/6250]  eta: 0:25:38  lr: 0.000041  grad: 0.1408 (0.1796)  loss: 0.8043 (0.7978)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [63]  [ 800/6250]  eta: 0:25:04  lr: 0.000041  grad: 0.1523 (0.1779)  loss: 0.7989 (0.7980)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [63]  [ 900/6250]  eta: 0:24:32  lr: 0.000041  grad: 0.1527 (0.1804)  loss: 0.8015 (0.7980)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [63]  [1000/6250]  eta: 0:24:02  lr: 0.000041  grad: 0.1630 (0.1806)  loss: 0.7939 (0.7980)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [63]  [1100/6250]  eta: 0:23:32  lr: 0.000041  grad: 0.1589 (0.1801)  loss: 0.7889 (0.7975)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [63]  [1200/6250]  eta: 0:23:02  lr: 0.000041  grad: 0.1538 (0.1807)  loss: 0.7913 (0.7971)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [63]  [1300/6250]  eta: 0:22:32  lr: 0.000041  grad: 0.1575 (0.1807)  loss: 0.7851 (0.7963)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [63]  [1400/6250]  eta: 0:22:03  lr: 0.000041  grad: 0.1616 (0.1817)  loss: 0.7836 (0.7955)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [63]  [1500/6250]  eta: 0:21:35  lr: 0.000041  grad: 0.1838 (0.1827)  loss: 0.7808 (0.7949)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [63]  [1600/6250]  eta: 0:21:06  lr: 0.000041  grad: 0.1626 (0.1829)  loss: 0.7897 (0.7944)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [63]  [1700/6250]  eta: 0:20:38  lr: 0.000041  grad: 0.1677 (0.1831)  loss: 0.7874 (0.7940)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [63]  [1800/6250]  eta: 0:20:09  lr: 0.000041  grad: 0.1577 (0.1831)  loss: 0.7837 (0.7937)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [63]  [1900/6250]  eta: 0:19:41  lr: 0.000041  grad: 0.1575 (0.1827)  loss: 0.7799 (0.7934)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [63]  [2000/6250]  eta: 0:19:13  lr: 0.000041  grad: 0.1591 (0.1835)  loss: 0.7906 (0.7930)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [63]  [2100/6250]  eta: 0:18:46  lr: 0.000041  grad: 0.1551 (0.1839)  loss: 0.7864 (0.7928)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [63]  [2200/6250]  eta: 0:18:18  lr: 0.000041  grad: 0.1934 (0.1849)  loss: 0.7837 (0.7925)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [63]  [2300/6250]  eta: 0:17:50  lr: 0.000041  grad: 0.1750 (0.1852)  loss: 0.7864 (0.7924)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [63]  [2400/6250]  eta: 0:17:23  lr: 0.000040  grad: 0.1624 (0.1852)  loss: 0.7828 (0.7921)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [63]  [2500/6250]  eta: 0:16:55  lr: 0.000040  grad: 0.1652 (0.1861)  loss: 0.7886 (0.7918)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [63]  [2600/6250]  eta: 0:16:28  lr: 0.000040  grad: 0.1819 (0.1866)  loss: 0.7839 (0.7916)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [63]  [2700/6250]  eta: 0:16:01  lr: 0.000040  grad: 0.1655 (0.1870)  loss: 0.7864 (0.7915)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [63]  [2800/6250]  eta: 0:15:33  lr: 0.000040  grad: 0.1711 (0.1874)  loss: 0.7910 (0.7914)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [63]  [2900/6250]  eta: 0:15:06  lr: 0.000040  grad: 0.1688 (0.1877)  loss: 0.7846 (0.7913)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [63]  [3000/6250]  eta: 0:14:38  lr: 0.000040  grad: 0.1527 (0.1881)  loss: 0.7784 (0.7912)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [63]  [3100/6250]  eta: 0:14:11  lr: 0.000040  grad: 0.1646 (0.1894)  loss: 0.7840 (0.7911)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [63]  [3200/6250]  eta: 0:13:44  lr: 0.000040  grad: 0.1573 (0.1896)  loss: 0.7881 (0.7910)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [63]  [3300/6250]  eta: 0:13:17  lr: 0.000040  grad: 0.1647 (0.1900)  loss: 0.7859 (0.7909)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [63]  [3400/6250]  eta: 0:12:49  lr: 0.000040  grad: 0.1534 (0.1896)  loss: 0.7861 (0.7908)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [63]  [3500/6250]  eta: 0:12:22  lr: 0.000040  grad: 0.1588 (0.1894)  loss: 0.7839 (0.7907)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [63]  [3600/6250]  eta: 0:11:55  lr: 0.000040  grad: 0.1406 (0.1898)  loss: 0.7917 (0.7907)  time: 0.2705  data: 0.0001  max mem: 26157
Train: [63]  [3700/6250]  eta: 0:11:28  lr: 0.000040  grad: 0.1638 (0.1896)  loss: 0.7938 (0.7907)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [63]  [3800/6250]  eta: 0:11:01  lr: 0.000040  grad: 0.1581 (0.1900)  loss: 0.7908 (0.7908)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [63]  [3900/6250]  eta: 0:10:34  lr: 0.000040  grad: 0.1497 (0.1902)  loss: 0.7834 (0.7907)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [63]  [4000/6250]  eta: 0:10:07  lr: 0.000040  grad: 0.1686 (0.1906)  loss: 0.7868 (0.7906)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [63]  [4100/6250]  eta: 0:09:40  lr: 0.000040  grad: 0.1586 (0.1909)  loss: 0.7894 (0.7906)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [63]  [4200/6250]  eta: 0:09:13  lr: 0.000040  grad: 0.1674 (0.1914)  loss: 0.7903 (0.7907)  time: 0.2722  data: 0.0002  max mem: 26157
Train: [63]  [4300/6250]  eta: 0:08:46  lr: 0.000040  grad: 0.1566 (0.1914)  loss: 0.7938 (0.7906)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [63]  [4400/6250]  eta: 0:08:19  lr: 0.000040  grad: 0.1503 (0.1926)  loss: 0.7878 (0.7906)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [63]  [4500/6250]  eta: 0:07:52  lr: 0.000040  grad: 0.1617 (0.1922)  loss: 0.7898 (0.7905)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [63]  [4600/6250]  eta: 0:07:25  lr: 0.000040  grad: 0.1514 (0.1928)  loss: 0.7937 (0.7905)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [63]  [4700/6250]  eta: 0:06:58  lr: 0.000040  grad: 0.1516 (0.1931)  loss: 0.7912 (0.7905)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [63]  [4800/6250]  eta: 0:06:31  lr: 0.000040  grad: 0.1642 (0.1930)  loss: 0.7906 (0.7905)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [63]  [4900/6250]  eta: 0:06:04  lr: 0.000040  grad: 0.1636 (0.1932)  loss: 0.7867 (0.7905)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [63]  [5000/6250]  eta: 0:05:37  lr: 0.000040  grad: 0.1487 (0.1929)  loss: 0.7900 (0.7905)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [63]  [5100/6250]  eta: 0:05:10  lr: 0.000040  grad: 0.1632 (0.1930)  loss: 0.7841 (0.7904)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [63]  [5200/6250]  eta: 0:04:43  lr: 0.000040  grad: 0.1525 (0.1928)  loss: 0.7913 (0.7904)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [63]  [5300/6250]  eta: 0:04:16  lr: 0.000040  grad: 0.1560 (0.1931)  loss: 0.7902 (0.7903)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [63]  [5400/6250]  eta: 0:03:49  lr: 0.000040  grad: 0.1527 (0.1928)  loss: 0.7825 (0.7903)  time: 0.2690  data: 0.0001  max mem: 26157
Train: [63]  [5500/6250]  eta: 0:03:22  lr: 0.000040  grad: 0.1687 (0.1929)  loss: 0.7870 (0.7903)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [63]  [5600/6250]  eta: 0:02:55  lr: 0.000039  grad: 0.1684 (0.1932)  loss: 0.7813 (0.7903)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [63]  [5700/6250]  eta: 0:02:28  lr: 0.000039  grad: 0.1621 (0.1932)  loss: 0.7857 (0.7902)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [63]  [5800/6250]  eta: 0:02:01  lr: 0.000039  grad: 0.1575 (0.1933)  loss: 0.7905 (0.7902)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [63]  [5900/6250]  eta: 0:01:34  lr: 0.000039  grad: 0.1579 (0.1930)  loss: 0.7882 (0.7901)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [63]  [6000/6250]  eta: 0:01:07  lr: 0.000039  grad: 0.1656 (0.1935)  loss: 0.7892 (0.7901)  time: 0.2709  data: 0.0001  max mem: 26157
Train: [63]  [6100/6250]  eta: 0:00:40  lr: 0.000039  grad: 0.1641 (0.1935)  loss: 0.7829 (0.7900)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [63]  [6200/6250]  eta: 0:00:13  lr: 0.000039  grad: 0.1593 (0.1937)  loss: 0.7888 (0.7900)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [63]  [6249/6250]  eta: 0:00:00  lr: 0.000039  grad: 0.1610 (0.1936)  loss: 0.7832 (0.7900)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [63] Total time: 0:28:12 (0.2708 s / it)
Averaged stats: lr: 0.000039  grad: 0.1610 (0.1936)  loss: 0.7832 (0.7900)
Eval (hcp-train-subset): [63]  [ 0/62]  eta: 0:04:04  loss: 0.8144 (0.8144)  time: 3.9418  data: 3.8364  max mem: 26157
Eval (hcp-train-subset): [63]  [61/62]  eta: 0:00:00  loss: 0.8140 (0.8150)  time: 0.1236  data: 0.0390  max mem: 26157
Eval (hcp-train-subset): [63] Total time: 0:00:12 (0.1990 s / it)
Averaged stats (hcp-train-subset): loss: 0.8140 (0.8150)
Making plots (hcp-train-subset): example=46
Eval (hcp-val): [63]  [ 0/62]  eta: 0:05:26  loss: 0.8214 (0.8214)  time: 5.2643  data: 5.1808  max mem: 26157
Eval (hcp-val): [63]  [61/62]  eta: 0:00:00  loss: 0.8237 (0.8249)  time: 0.1085  data: 0.0240  max mem: 26157
Eval (hcp-val): [63] Total time: 0:00:12 (0.1981 s / it)
Averaged stats (hcp-val): loss: 0.8237 (0.8249)
Making plots (hcp-val): example=46
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [64]  [   0/6250]  eta: 9:55:47  lr: 0.000039  grad: 0.1823 (0.1823)  loss: 0.8267 (0.8267)  time: 5.7195  data: 5.4430  max mem: 26157
Train: [64]  [ 100/6250]  eta: 0:33:09  lr: 0.000039  grad: 0.1599 (0.2203)  loss: 0.8048 (0.8060)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [64]  [ 200/6250]  eta: 0:29:52  lr: 0.000039  grad: 0.1700 (0.2087)  loss: 0.7896 (0.7987)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [64]  [ 300/6250]  eta: 0:28:28  lr: 0.000039  grad: 0.1608 (0.1968)  loss: 0.7927 (0.7987)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [64]  [ 400/6250]  eta: 0:27:32  lr: 0.000039  grad: 0.1354 (0.1902)  loss: 0.8010 (0.7980)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [64]  [ 500/6250]  eta: 0:26:48  lr: 0.000039  grad: 0.1497 (0.1863)  loss: 0.7939 (0.7977)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [64]  [ 600/6250]  eta: 0:26:09  lr: 0.000039  grad: 0.1549 (0.1828)  loss: 0.7987 (0.7981)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [64]  [ 700/6250]  eta: 0:25:34  lr: 0.000039  grad: 0.1586 (0.1829)  loss: 0.7960 (0.7977)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [64]  [ 800/6250]  eta: 0:25:06  lr: 0.000039  grad: 0.1480 (0.1830)  loss: 0.7909 (0.7969)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [64]  [ 900/6250]  eta: 0:24:34  lr: 0.000039  grad: 0.1493 (0.1832)  loss: 0.7996 (0.7964)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [64]  [1000/6250]  eta: 0:24:03  lr: 0.000039  grad: 0.1351 (0.1818)  loss: 0.7856 (0.7957)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [64]  [1100/6250]  eta: 0:23:32  lr: 0.000039  grad: 0.1531 (0.1811)  loss: 0.7895 (0.7950)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [64]  [1200/6250]  eta: 0:23:02  lr: 0.000039  grad: 0.1523 (0.1811)  loss: 0.7879 (0.7944)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [64]  [1300/6250]  eta: 0:22:33  lr: 0.000039  grad: 0.1613 (0.1810)  loss: 0.7947 (0.7941)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [64]  [1400/6250]  eta: 0:22:04  lr: 0.000039  grad: 0.1366 (0.1800)  loss: 0.7913 (0.7938)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [64]  [1500/6250]  eta: 0:21:35  lr: 0.000039  grad: 0.1414 (0.1804)  loss: 0.7809 (0.7935)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [64]  [1600/6250]  eta: 0:21:07  lr: 0.000039  grad: 0.1522 (0.1803)  loss: 0.7918 (0.7931)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [64]  [1700/6250]  eta: 0:20:39  lr: 0.000039  grad: 0.1447 (0.1809)  loss: 0.7863 (0.7928)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [64]  [1800/6250]  eta: 0:20:11  lr: 0.000039  grad: 0.1668 (0.1826)  loss: 0.7838 (0.7924)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [64]  [1900/6250]  eta: 0:19:43  lr: 0.000039  grad: 0.1445 (0.1829)  loss: 0.7898 (0.7921)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [64]  [2000/6250]  eta: 0:19:15  lr: 0.000039  grad: 0.1437 (0.1843)  loss: 0.7972 (0.7919)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [64]  [2100/6250]  eta: 0:18:47  lr: 0.000039  grad: 0.1646 (0.1839)  loss: 0.7820 (0.7916)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [64]  [2200/6250]  eta: 0:18:19  lr: 0.000039  grad: 0.1467 (0.1829)  loss: 0.7863 (0.7914)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [64]  [2300/6250]  eta: 0:17:52  lr: 0.000039  grad: 0.1498 (0.1835)  loss: 0.7952 (0.7914)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [64]  [2400/6250]  eta: 0:17:24  lr: 0.000039  grad: 0.1562 (0.1842)  loss: 0.7938 (0.7912)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [64]  [2500/6250]  eta: 0:16:57  lr: 0.000039  grad: 0.1679 (0.1862)  loss: 0.7801 (0.7910)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [64]  [2600/6250]  eta: 0:16:29  lr: 0.000039  grad: 0.1536 (0.1860)  loss: 0.7880 (0.7909)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [64]  [2700/6250]  eta: 0:16:02  lr: 0.000038  grad: 0.1512 (0.1864)  loss: 0.7868 (0.7907)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [64]  [2800/6250]  eta: 0:15:34  lr: 0.000038  grad: 0.1495 (0.1868)  loss: 0.7880 (0.7906)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [64]  [2900/6250]  eta: 0:15:07  lr: 0.000038  grad: 0.1495 (0.1862)  loss: 0.7879 (0.7904)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [64]  [3000/6250]  eta: 0:14:40  lr: 0.000038  grad: 0.1533 (0.1861)  loss: 0.7859 (0.7903)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [64]  [3100/6250]  eta: 0:14:12  lr: 0.000038  grad: 0.1581 (0.1860)  loss: 0.7848 (0.7903)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [64]  [3200/6250]  eta: 0:13:45  lr: 0.000038  grad: 0.1524 (0.1862)  loss: 0.7804 (0.7902)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [64]  [3300/6250]  eta: 0:13:18  lr: 0.000038  grad: 0.1479 (0.1868)  loss: 0.7876 (0.7901)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [64]  [3400/6250]  eta: 0:12:50  lr: 0.000038  grad: 0.1481 (0.1876)  loss: 0.7882 (0.7900)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [64]  [3500/6250]  eta: 0:12:23  lr: 0.000038  grad: 0.1524 (0.1877)  loss: 0.7883 (0.7898)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [64]  [3600/6250]  eta: 0:11:56  lr: 0.000038  grad: 0.1509 (0.1882)  loss: 0.7896 (0.7896)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [64]  [3700/6250]  eta: 0:11:29  lr: 0.000038  grad: 0.1544 (0.1890)  loss: 0.7890 (0.7896)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [64]  [3800/6250]  eta: 0:11:02  lr: 0.000038  grad: 0.1572 (0.1891)  loss: 0.7916 (0.7896)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [64]  [3900/6250]  eta: 0:10:35  lr: 0.000038  grad: 0.1538 (0.1898)  loss: 0.7863 (0.7896)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [64]  [4000/6250]  eta: 0:10:08  lr: 0.000038  grad: 0.1528 (0.1899)  loss: 0.7883 (0.7896)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [64]  [4100/6250]  eta: 0:09:40  lr: 0.000038  grad: 0.1556 (0.1909)  loss: 0.7869 (0.7896)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [64]  [4200/6250]  eta: 0:09:13  lr: 0.000038  grad: 0.1547 (0.1914)  loss: 0.7885 (0.7896)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [64]  [4300/6250]  eta: 0:08:46  lr: 0.000038  grad: 0.1724 (0.1916)  loss: 0.7837 (0.7896)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [64]  [4400/6250]  eta: 0:08:19  lr: 0.000038  grad: 0.1735 (0.1921)  loss: 0.7888 (0.7895)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [64]  [4500/6250]  eta: 0:07:52  lr: 0.000038  grad: 0.1567 (0.1924)  loss: 0.7862 (0.7895)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [64]  [4600/6250]  eta: 0:07:25  lr: 0.000038  grad: 0.1578 (0.1921)  loss: 0.7842 (0.7894)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [64]  [4700/6250]  eta: 0:06:58  lr: 0.000038  grad: 0.1806 (0.1929)  loss: 0.7785 (0.7893)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [64]  [4800/6250]  eta: 0:06:31  lr: 0.000038  grad: 0.1700 (0.1936)  loss: 0.7816 (0.7892)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [64]  [4900/6250]  eta: 0:06:04  lr: 0.000038  grad: 0.1572 (0.1934)  loss: 0.7836 (0.7890)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [64]  [5000/6250]  eta: 0:05:37  lr: 0.000038  grad: 0.1546 (0.1934)  loss: 0.7829 (0.7888)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [64]  [5100/6250]  eta: 0:05:10  lr: 0.000038  grad: 0.1667 (0.1937)  loss: 0.7798 (0.7887)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [64]  [5200/6250]  eta: 0:04:43  lr: 0.000038  grad: 0.1474 (0.1936)  loss: 0.7877 (0.7886)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [64]  [5300/6250]  eta: 0:04:16  lr: 0.000038  grad: 0.1766 (0.1938)  loss: 0.7768 (0.7885)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [64]  [5400/6250]  eta: 0:03:49  lr: 0.000038  grad: 0.1672 (0.1937)  loss: 0.7862 (0.7884)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [64]  [5500/6250]  eta: 0:03:22  lr: 0.000038  grad: 0.1532 (0.1939)  loss: 0.7892 (0.7884)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [64]  [5600/6250]  eta: 0:02:55  lr: 0.000038  grad: 0.1681 (0.1937)  loss: 0.7866 (0.7883)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [64]  [5700/6250]  eta: 0:02:28  lr: 0.000038  grad: 0.1629 (0.1935)  loss: 0.7824 (0.7883)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [64]  [5800/6250]  eta: 0:02:01  lr: 0.000038  grad: 0.1668 (0.1940)  loss: 0.7892 (0.7883)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [64]  [5900/6250]  eta: 0:01:34  lr: 0.000037  grad: 0.1619 (0.1941)  loss: 0.7855 (0.7883)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [64]  [6000/6250]  eta: 0:01:07  lr: 0.000037  grad: 0.1836 (0.1948)  loss: 0.7854 (0.7883)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [64]  [6100/6250]  eta: 0:00:40  lr: 0.000037  grad: 0.1607 (0.1954)  loss: 0.7822 (0.7882)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [64]  [6200/6250]  eta: 0:00:13  lr: 0.000037  grad: 0.1789 (0.1961)  loss: 0.7869 (0.7882)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [64]  [6249/6250]  eta: 0:00:00  lr: 0.000037  grad: 0.1501 (0.1961)  loss: 0.7840 (0.7882)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [64] Total time: 0:28:11 (0.2706 s / it)
Averaged stats: lr: 0.000037  grad: 0.1501 (0.1961)  loss: 0.7840 (0.7882)
Eval (hcp-train-subset): [64]  [ 0/62]  eta: 0:05:03  loss: 0.8213 (0.8213)  time: 4.8912  data: 4.8076  max mem: 26157
Eval (hcp-train-subset): [64]  [61/62]  eta: 0:00:00  loss: 0.8102 (0.8127)  time: 0.1176  data: 0.0346  max mem: 26157
Eval (hcp-train-subset): [64] Total time: 0:00:11 (0.1889 s / it)
Averaged stats (hcp-train-subset): loss: 0.8102 (0.8127)
Making plots (hcp-train-subset): example=10
Eval (hcp-val): [64]  [ 0/62]  eta: 0:03:20  loss: 0.8227 (0.8227)  time: 3.2347  data: 3.1213  max mem: 26157
Eval (hcp-val): [64]  [61/62]  eta: 0:00:00  loss: 0.8242 (0.8254)  time: 0.1170  data: 0.0346  max mem: 26157
Eval (hcp-val): [64] Total time: 0:00:11 (0.1872 s / it)
Averaged stats (hcp-val): loss: 0.8242 (0.8254)
Making plots (hcp-val): example=32
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [65]  [   0/6250]  eta: 6:45:55  lr: 0.000037  grad: 0.1111 (0.1111)  loss: 0.8515 (0.8515)  time: 3.8969  data: 3.5636  max mem: 26157
Train: [65]  [ 100/6250]  eta: 0:32:29  lr: 0.000037  grad: 0.1761 (0.2166)  loss: 0.8105 (0.8126)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [65]  [ 200/6250]  eta: 0:29:32  lr: 0.000037  grad: 0.1461 (0.2118)  loss: 0.8143 (0.8101)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [65]  [ 300/6250]  eta: 0:28:14  lr: 0.000037  grad: 0.1629 (0.1991)  loss: 0.7965 (0.8059)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [65]  [ 400/6250]  eta: 0:27:22  lr: 0.000037  grad: 0.1773 (0.1957)  loss: 0.7875 (0.8019)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [65]  [ 500/6250]  eta: 0:26:40  lr: 0.000037  grad: 0.1557 (0.1940)  loss: 0.7952 (0.7997)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [65]  [ 600/6250]  eta: 0:26:03  lr: 0.000037  grad: 0.1730 (0.1938)  loss: 0.7884 (0.7977)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [65]  [ 700/6250]  eta: 0:25:29  lr: 0.000037  grad: 0.1602 (0.1941)  loss: 0.7866 (0.7959)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [65]  [ 800/6250]  eta: 0:24:57  lr: 0.000037  grad: 0.1528 (0.1933)  loss: 0.7862 (0.7949)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [65]  [ 900/6250]  eta: 0:24:26  lr: 0.000037  grad: 0.1575 (0.1907)  loss: 0.7855 (0.7943)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [65]  [1000/6250]  eta: 0:23:56  lr: 0.000037  grad: 0.1489 (0.1905)  loss: 0.7896 (0.7941)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [65]  [1100/6250]  eta: 0:23:26  lr: 0.000037  grad: 0.1719 (0.1909)  loss: 0.7901 (0.7936)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [65]  [1200/6250]  eta: 0:22:57  lr: 0.000037  grad: 0.1577 (0.1922)  loss: 0.7855 (0.7931)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [65]  [1300/6250]  eta: 0:22:28  lr: 0.000037  grad: 0.1627 (0.1951)  loss: 0.7913 (0.7925)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [65]  [1400/6250]  eta: 0:22:00  lr: 0.000037  grad: 0.1571 (0.1967)  loss: 0.7819 (0.7919)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [65]  [1500/6250]  eta: 0:21:31  lr: 0.000037  grad: 0.1562 (0.1962)  loss: 0.7887 (0.7916)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [65]  [1600/6250]  eta: 0:21:03  lr: 0.000037  grad: 0.1556 (0.1958)  loss: 0.7871 (0.7911)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [65]  [1700/6250]  eta: 0:20:35  lr: 0.000037  grad: 0.1636 (0.1957)  loss: 0.7929 (0.7909)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [65]  [1800/6250]  eta: 0:20:07  lr: 0.000037  grad: 0.1795 (0.1966)  loss: 0.7820 (0.7904)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [65]  [1900/6250]  eta: 0:19:39  lr: 0.000037  grad: 0.1530 (0.1982)  loss: 0.7841 (0.7902)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [65]  [2000/6250]  eta: 0:19:11  lr: 0.000037  grad: 0.1564 (0.1984)  loss: 0.7921 (0.7901)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [65]  [2100/6250]  eta: 0:18:44  lr: 0.000037  grad: 0.1580 (0.2021)  loss: 0.7869 (0.7900)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [65]  [2200/6250]  eta: 0:18:16  lr: 0.000037  grad: 0.1694 (0.2045)  loss: 0.7905 (0.7898)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [65]  [2300/6250]  eta: 0:17:49  lr: 0.000037  grad: 0.2151 (0.2088)  loss: 0.7899 (0.7898)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [65]  [2400/6250]  eta: 0:17:21  lr: 0.000037  grad: 0.1739 (0.2093)  loss: 0.7895 (0.7897)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [65]  [2500/6250]  eta: 0:16:54  lr: 0.000037  grad: 0.1672 (0.2087)  loss: 0.7873 (0.7896)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [65]  [2600/6250]  eta: 0:16:26  lr: 0.000037  grad: 0.1686 (0.2086)  loss: 0.7861 (0.7894)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [65]  [2700/6250]  eta: 0:15:59  lr: 0.000037  grad: 0.1584 (0.2077)  loss: 0.7909 (0.7894)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [65]  [2800/6250]  eta: 0:15:32  lr: 0.000037  grad: 0.1714 (0.2071)  loss: 0.7843 (0.7892)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [65]  [2900/6250]  eta: 0:15:05  lr: 0.000037  grad: 0.1604 (0.2079)  loss: 0.7836 (0.7891)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [65]  [3000/6250]  eta: 0:14:37  lr: 0.000036  grad: 0.1744 (0.2088)  loss: 0.7907 (0.7890)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [65]  [3100/6250]  eta: 0:14:10  lr: 0.000036  grad: 0.1608 (0.2086)  loss: 0.7885 (0.7888)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [65]  [3200/6250]  eta: 0:13:43  lr: 0.000036  grad: 0.1661 (0.2087)  loss: 0.7847 (0.7887)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [65]  [3300/6250]  eta: 0:13:16  lr: 0.000036  grad: 0.1650 (0.2089)  loss: 0.7881 (0.7884)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [65]  [3400/6250]  eta: 0:12:49  lr: 0.000036  grad: 0.1536 (0.2089)  loss: 0.7869 (0.7883)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [65]  [3500/6250]  eta: 0:12:22  lr: 0.000036  grad: 0.1513 (0.2109)  loss: 0.7945 (0.7882)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [65]  [3600/6250]  eta: 0:11:55  lr: 0.000036  grad: 0.1769 (0.2109)  loss: 0.7809 (0.7881)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [65]  [3700/6250]  eta: 0:11:28  lr: 0.000036  grad: 0.1708 (0.2116)  loss: 0.7823 (0.7881)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [65]  [3800/6250]  eta: 0:11:00  lr: 0.000036  grad: 0.1604 (0.2120)  loss: 0.7859 (0.7880)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [65]  [3900/6250]  eta: 0:10:33  lr: 0.000036  grad: 0.1672 (0.2128)  loss: 0.7810 (0.7879)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [65]  [4000/6250]  eta: 0:10:06  lr: 0.000036  grad: 0.1688 (0.2137)  loss: 0.7801 (0.7877)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [65]  [4100/6250]  eta: 0:09:39  lr: 0.000036  grad: 0.1626 (0.2129)  loss: 0.7828 (0.7877)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [65]  [4200/6250]  eta: 0:09:12  lr: 0.000036  grad: 0.1560 (0.2130)  loss: 0.7807 (0.7876)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [65]  [4300/6250]  eta: 0:08:45  lr: 0.000036  grad: 0.1655 (0.2126)  loss: 0.7784 (0.7876)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [65]  [4400/6250]  eta: 0:08:18  lr: 0.000036  grad: 0.1564 (0.2125)  loss: 0.7995 (0.7876)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [65]  [4500/6250]  eta: 0:07:51  lr: 0.000036  grad: 0.1716 (0.2131)  loss: 0.7861 (0.7876)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [65]  [4600/6250]  eta: 0:07:24  lr: 0.000036  grad: 0.1585 (0.2131)  loss: 0.7836 (0.7876)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [65]  [4700/6250]  eta: 0:06:57  lr: 0.000036  grad: 0.1635 (0.2128)  loss: 0.7901 (0.7876)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [65]  [4800/6250]  eta: 0:06:30  lr: 0.000036  grad: 0.1681 (0.2135)  loss: 0.7929 (0.7876)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [65]  [4900/6250]  eta: 0:06:03  lr: 0.000036  grad: 0.1689 (0.2135)  loss: 0.7863 (0.7876)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [65]  [5000/6250]  eta: 0:05:36  lr: 0.000036  grad: 0.1671 (0.2139)  loss: 0.7786 (0.7876)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [65]  [5100/6250]  eta: 0:05:09  lr: 0.000036  grad: 0.1577 (0.2152)  loss: 0.7832 (0.7875)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [65]  [5200/6250]  eta: 0:04:42  lr: 0.000036  grad: 0.1709 (0.2159)  loss: 0.7779 (0.7874)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [65]  [5300/6250]  eta: 0:04:16  lr: 0.000036  grad: 0.1706 (0.2163)  loss: 0.7848 (0.7873)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [65]  [5400/6250]  eta: 0:03:49  lr: 0.000036  grad: 0.1815 (0.2167)  loss: 0.7854 (0.7873)  time: 0.2710  data: 0.0002  max mem: 26157
Train: [65]  [5500/6250]  eta: 0:03:22  lr: 0.000036  grad: 0.1688 (0.2180)  loss: 0.7929 (0.7873)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [65]  [5600/6250]  eta: 0:02:55  lr: 0.000036  grad: 0.1743 (0.2183)  loss: 0.7950 (0.7873)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [65]  [5700/6250]  eta: 0:02:28  lr: 0.000036  grad: 0.1525 (0.2191)  loss: 0.7912 (0.7873)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [65]  [5800/6250]  eta: 0:02:01  lr: 0.000036  grad: 0.1844 (0.2192)  loss: 0.7857 (0.7874)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [65]  [5900/6250]  eta: 0:01:34  lr: 0.000036  grad: 0.1708 (0.2200)  loss: 0.7924 (0.7874)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [65]  [6000/6250]  eta: 0:01:07  lr: 0.000036  grad: 0.1657 (0.2198)  loss: 0.7913 (0.7874)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [65]  [6100/6250]  eta: 0:00:40  lr: 0.000036  grad: 0.1661 (0.2203)  loss: 0.7865 (0.7874)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [65]  [6200/6250]  eta: 0:00:13  lr: 0.000036  grad: 0.1640 (0.2200)  loss: 0.7829 (0.7874)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [65]  [6249/6250]  eta: 0:00:00  lr: 0.000036  grad: 0.1959 (0.2200)  loss: 0.7849 (0.7874)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [65] Total time: 0:28:09 (0.2703 s / it)
Averaged stats: lr: 0.000036  grad: 0.1959 (0.2200)  loss: 0.7849 (0.7874)
Eval (hcp-train-subset): [65]  [ 0/62]  eta: 0:05:29  loss: 0.8203 (0.8203)  time: 5.3127  data: 5.2298  max mem: 26157
Eval (hcp-train-subset): [65]  [61/62]  eta: 0:00:00  loss: 0.8065 (0.8102)  time: 0.1312  data: 0.0468  max mem: 26157
Eval (hcp-train-subset): [65] Total time: 0:00:12 (0.1983 s / it)
Averaged stats (hcp-train-subset): loss: 0.8065 (0.8102)
Making plots (hcp-train-subset): example=47
Eval (hcp-val): [65]  [ 0/62]  eta: 0:05:39  loss: 0.8201 (0.8201)  time: 5.4812  data: 5.3974  max mem: 26157
Eval (hcp-val): [65]  [61/62]  eta: 0:00:00  loss: 0.8236 (0.8248)  time: 0.1093  data: 0.0244  max mem: 26157
Eval (hcp-val): [65] Total time: 0:00:12 (0.1975 s / it)
Averaged stats (hcp-val): loss: 0.8236 (0.8248)
Making plots (hcp-val): example=59
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [66]  [   0/6250]  eta: 9:36:42  lr: 0.000036  grad: 0.2856 (0.2856)  loss: 0.8051 (0.8051)  time: 5.5364  data: 5.2576  max mem: 26157
Train: [66]  [ 100/6250]  eta: 0:33:29  lr: 0.000035  grad: 0.1727 (0.2341)  loss: 0.8104 (0.8120)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [66]  [ 200/6250]  eta: 0:30:00  lr: 0.000035  grad: 0.1688 (0.2296)  loss: 0.7939 (0.8038)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [66]  [ 300/6250]  eta: 0:28:32  lr: 0.000035  grad: 0.1729 (0.2353)  loss: 0.7840 (0.7984)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [66]  [ 400/6250]  eta: 0:27:35  lr: 0.000035  grad: 0.1723 (0.2393)  loss: 0.7918 (0.7958)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [66]  [ 500/6250]  eta: 0:26:50  lr: 0.000035  grad: 0.1952 (0.2395)  loss: 0.7890 (0.7930)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [66]  [ 600/6250]  eta: 0:26:10  lr: 0.000035  grad: 0.1785 (0.2420)  loss: 0.7874 (0.7915)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [66]  [ 700/6250]  eta: 0:25:36  lr: 0.000035  grad: 0.1939 (0.2380)  loss: 0.7836 (0.7904)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [66]  [ 800/6250]  eta: 0:25:03  lr: 0.000035  grad: 0.1752 (0.2376)  loss: 0.7909 (0.7894)  time: 0.2720  data: 0.0002  max mem: 26157
Train: [66]  [ 900/6250]  eta: 0:24:31  lr: 0.000035  grad: 0.1581 (0.2340)  loss: 0.7886 (0.7887)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [66]  [1000/6250]  eta: 0:24:00  lr: 0.000035  grad: 0.1703 (0.2358)  loss: 0.7769 (0.7886)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [66]  [1100/6250]  eta: 0:23:31  lr: 0.000035  grad: 0.1752 (0.2355)  loss: 0.7869 (0.7885)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [66]  [1200/6250]  eta: 0:23:01  lr: 0.000035  grad: 0.1593 (0.2328)  loss: 0.7756 (0.7881)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [66]  [1300/6250]  eta: 0:22:32  lr: 0.000035  grad: 0.1638 (0.2295)  loss: 0.7731 (0.7878)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [66]  [1400/6250]  eta: 0:22:03  lr: 0.000035  grad: 0.1691 (0.2263)  loss: 0.7838 (0.7876)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [66]  [1500/6250]  eta: 0:21:35  lr: 0.000035  grad: 0.1687 (0.2250)  loss: 0.7803 (0.7873)  time: 0.2713  data: 0.0002  max mem: 26157
Train: [66]  [1600/6250]  eta: 0:21:07  lr: 0.000035  grad: 0.1532 (0.2242)  loss: 0.7825 (0.7871)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [66]  [1700/6250]  eta: 0:20:38  lr: 0.000035  grad: 0.1681 (0.2236)  loss: 0.7741 (0.7869)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [66]  [1800/6250]  eta: 0:20:10  lr: 0.000035  grad: 0.1589 (0.2241)  loss: 0.7878 (0.7868)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [66]  [1900/6250]  eta: 0:19:42  lr: 0.000035  grad: 0.1671 (0.2253)  loss: 0.7913 (0.7867)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [66]  [2000/6250]  eta: 0:19:15  lr: 0.000035  grad: 0.1535 (0.2249)  loss: 0.7843 (0.7865)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [66]  [2100/6250]  eta: 0:18:47  lr: 0.000035  grad: 0.1702 (0.2286)  loss: 0.7836 (0.7863)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [66]  [2200/6250]  eta: 0:18:19  lr: 0.000035  grad: 0.1639 (0.2281)  loss: 0.7908 (0.7863)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [66]  [2300/6250]  eta: 0:17:52  lr: 0.000035  grad: 0.1734 (0.2275)  loss: 0.7807 (0.7862)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [66]  [2400/6250]  eta: 0:17:24  lr: 0.000035  grad: 0.1661 (0.2285)  loss: 0.7831 (0.7859)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [66]  [2500/6250]  eta: 0:16:57  lr: 0.000035  grad: 0.1579 (0.2287)  loss: 0.7892 (0.7858)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [66]  [2600/6250]  eta: 0:16:29  lr: 0.000035  grad: 0.1569 (0.2276)  loss: 0.7823 (0.7858)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [66]  [2700/6250]  eta: 0:16:02  lr: 0.000035  grad: 0.1578 (0.2281)  loss: 0.7866 (0.7857)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [66]  [2800/6250]  eta: 0:15:35  lr: 0.000035  grad: 0.1554 (0.2280)  loss: 0.7886 (0.7857)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [66]  [2900/6250]  eta: 0:15:07  lr: 0.000035  grad: 0.1573 (0.2266)  loss: 0.7829 (0.7857)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [66]  [3000/6250]  eta: 0:14:40  lr: 0.000035  grad: 0.1620 (0.2258)  loss: 0.7825 (0.7857)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [66]  [3100/6250]  eta: 0:14:12  lr: 0.000035  grad: 0.1562 (0.2250)  loss: 0.7914 (0.7857)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [66]  [3200/6250]  eta: 0:13:45  lr: 0.000035  grad: 0.1597 (0.2284)  loss: 0.7876 (0.7858)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [66]  [3300/6250]  eta: 0:13:18  lr: 0.000035  grad: 0.1543 (0.2277)  loss: 0.7825 (0.7859)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [66]  [3400/6250]  eta: 0:12:51  lr: 0.000035  grad: 0.1573 (0.2292)  loss: 0.7998 (0.7860)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [66]  [3500/6250]  eta: 0:12:23  lr: 0.000034  grad: 0.1525 (0.2287)  loss: 0.7978 (0.7861)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [66]  [3600/6250]  eta: 0:11:56  lr: 0.000034  grad: 0.1627 (0.2282)  loss: 0.7868 (0.7862)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [66]  [3700/6250]  eta: 0:11:29  lr: 0.000034  grad: 0.1592 (0.2274)  loss: 0.7896 (0.7863)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [66]  [3800/6250]  eta: 0:11:02  lr: 0.000034  grad: 0.1688 (0.2271)  loss: 0.7896 (0.7863)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [66]  [3900/6250]  eta: 0:10:35  lr: 0.000034  grad: 0.1765 (0.2278)  loss: 0.7806 (0.7864)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [66]  [4000/6250]  eta: 0:10:08  lr: 0.000034  grad: 0.1622 (0.2292)  loss: 0.7844 (0.7864)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [66]  [4100/6250]  eta: 0:09:40  lr: 0.000034  grad: 0.1632 (0.2294)  loss: 0.7916 (0.7865)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [66]  [4200/6250]  eta: 0:09:13  lr: 0.000034  grad: 0.1629 (0.2292)  loss: 0.7943 (0.7866)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [66]  [4300/6250]  eta: 0:08:46  lr: 0.000034  grad: 0.1552 (0.2298)  loss: 0.7870 (0.7867)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [66]  [4400/6250]  eta: 0:08:19  lr: 0.000034  grad: 0.1576 (0.2297)  loss: 0.7883 (0.7868)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [66]  [4500/6250]  eta: 0:07:52  lr: 0.000034  grad: 0.1636 (0.2296)  loss: 0.7945 (0.7869)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [66]  [4600/6250]  eta: 0:07:25  lr: 0.000034  grad: 0.1742 (0.2289)  loss: 0.7799 (0.7870)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [66]  [4700/6250]  eta: 0:06:58  lr: 0.000034  grad: 0.1731 (0.2283)  loss: 0.7831 (0.7870)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [66]  [4800/6250]  eta: 0:06:31  lr: 0.000034  grad: 0.1632 (0.2293)  loss: 0.7907 (0.7870)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [66]  [4900/6250]  eta: 0:06:04  lr: 0.000034  grad: 0.1566 (0.2289)  loss: 0.7882 (0.7870)  time: 0.2709  data: 0.0002  max mem: 26157
Train: [66]  [5000/6250]  eta: 0:05:37  lr: 0.000034  grad: 0.1770 (0.2296)  loss: 0.7814 (0.7869)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [66]  [5100/6250]  eta: 0:05:10  lr: 0.000034  grad: 0.1640 (0.2299)  loss: 0.7832 (0.7869)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [66]  [5200/6250]  eta: 0:04:43  lr: 0.000034  grad: 0.1661 (0.2295)  loss: 0.7788 (0.7868)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [66]  [5300/6250]  eta: 0:04:16  lr: 0.000034  grad: 0.1621 (0.2292)  loss: 0.7884 (0.7868)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [66]  [5400/6250]  eta: 0:03:49  lr: 0.000034  grad: 0.1677 (0.2292)  loss: 0.7830 (0.7868)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [66]  [5500/6250]  eta: 0:03:22  lr: 0.000034  grad: 0.1686 (0.2295)  loss: 0.7864 (0.7867)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [66]  [5600/6250]  eta: 0:02:55  lr: 0.000034  grad: 0.1704 (0.2297)  loss: 0.7834 (0.7867)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [66]  [5700/6250]  eta: 0:02:28  lr: 0.000034  grad: 0.1652 (0.2291)  loss: 0.7756 (0.7866)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [66]  [5800/6250]  eta: 0:02:01  lr: 0.000034  grad: 0.1783 (0.2292)  loss: 0.7801 (0.7866)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [66]  [5900/6250]  eta: 0:01:34  lr: 0.000034  grad: 0.1550 (0.2292)  loss: 0.7762 (0.7865)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [66]  [6000/6250]  eta: 0:01:07  lr: 0.000034  grad: 0.1598 (0.2289)  loss: 0.7836 (0.7865)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [66]  [6100/6250]  eta: 0:00:40  lr: 0.000034  grad: 0.1657 (0.2289)  loss: 0.7799 (0.7864)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [66]  [6200/6250]  eta: 0:00:13  lr: 0.000034  grad: 0.1673 (0.2288)  loss: 0.7711 (0.7863)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [66]  [6249/6250]  eta: 0:00:00  lr: 0.000034  grad: 0.1642 (0.2286)  loss: 0.7913 (0.7863)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [66] Total time: 0:28:10 (0.2705 s / it)
Averaged stats: lr: 0.000034  grad: 0.1642 (0.2286)  loss: 0.7913 (0.7863)
Eval (hcp-train-subset): [66]  [ 0/62]  eta: 0:03:20  loss: 0.8197 (0.8197)  time: 3.2305  data: 3.1253  max mem: 26157
Eval (hcp-train-subset): [66]  [61/62]  eta: 0:00:00  loss: 0.8101 (0.8119)  time: 0.1094  data: 0.0266  max mem: 26157
Eval (hcp-train-subset): [66] Total time: 0:00:12 (0.1944 s / it)
Averaged stats (hcp-train-subset): loss: 0.8101 (0.8119)
Making plots (hcp-train-subset): example=16
Eval (hcp-val): [66]  [ 0/62]  eta: 0:03:42  loss: 0.8205 (0.8205)  time: 3.5948  data: 3.5027  max mem: 26157
Eval (hcp-val): [66]  [61/62]  eta: 0:00:00  loss: 0.8243 (0.8254)  time: 0.1193  data: 0.0365  max mem: 26157
Eval (hcp-val): [66] Total time: 0:00:12 (0.1984 s / it)
Averaged stats (hcp-val): loss: 0.8243 (0.8254)
Making plots (hcp-val): example=16
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [67]  [   0/6250]  eta: 7:48:41  lr: 0.000034  grad: 0.3304 (0.3304)  loss: 0.7778 (0.7778)  time: 4.4995  data: 4.1973  max mem: 26157
Train: [67]  [ 100/6250]  eta: 0:32:33  lr: 0.000034  grad: 0.1899 (0.2479)  loss: 0.8115 (0.8145)  time: 0.2699  data: 0.0002  max mem: 26157
Train: [67]  [ 200/6250]  eta: 0:29:40  lr: 0.000034  grad: 0.1862 (0.2402)  loss: 0.7933 (0.8051)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [67]  [ 300/6250]  eta: 0:28:19  lr: 0.000034  grad: 0.1761 (0.2278)  loss: 0.7849 (0.7984)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [67]  [ 400/6250]  eta: 0:27:25  lr: 0.000034  grad: 0.1711 (0.2243)  loss: 0.7781 (0.7957)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [67]  [ 500/6250]  eta: 0:26:42  lr: 0.000034  grad: 0.1690 (0.2281)  loss: 0.7865 (0.7937)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [67]  [ 600/6250]  eta: 0:26:04  lr: 0.000033  grad: 0.2022 (0.2327)  loss: 0.7793 (0.7916)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [67]  [ 700/6250]  eta: 0:25:29  lr: 0.000033  grad: 0.1653 (0.2304)  loss: 0.7847 (0.7902)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [67]  [ 800/6250]  eta: 0:24:57  lr: 0.000033  grad: 0.1677 (0.2277)  loss: 0.7923 (0.7897)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [67]  [ 900/6250]  eta: 0:24:26  lr: 0.000033  grad: 0.1839 (0.2293)  loss: 0.7862 (0.7896)  time: 0.2704  data: 0.0002  max mem: 26157
Train: [67]  [1000/6250]  eta: 0:23:57  lr: 0.000033  grad: 0.1570 (0.2262)  loss: 0.7884 (0.7896)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [67]  [1100/6250]  eta: 0:23:27  lr: 0.000033  grad: 0.1666 (0.2305)  loss: 0.7832 (0.7892)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [67]  [1200/6250]  eta: 0:22:57  lr: 0.000033  grad: 0.1712 (0.2306)  loss: 0.7838 (0.7890)  time: 0.2682  data: 0.0003  max mem: 26157
Train: [67]  [1300/6250]  eta: 0:22:28  lr: 0.000033  grad: 0.1670 (0.2279)  loss: 0.7816 (0.7886)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [67]  [1400/6250]  eta: 0:21:59  lr: 0.000033  grad: 0.1669 (0.2290)  loss: 0.7774 (0.7882)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [67]  [1500/6250]  eta: 0:21:31  lr: 0.000033  grad: 0.1622 (0.2268)  loss: 0.7865 (0.7881)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [67]  [1600/6250]  eta: 0:21:03  lr: 0.000033  grad: 0.1606 (0.2273)  loss: 0.7890 (0.7878)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [67]  [1700/6250]  eta: 0:20:35  lr: 0.000033  grad: 0.1690 (0.2250)  loss: 0.7821 (0.7875)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [67]  [1800/6250]  eta: 0:20:07  lr: 0.000033  grad: 0.1572 (0.2244)  loss: 0.7841 (0.7875)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [67]  [1900/6250]  eta: 0:19:39  lr: 0.000033  grad: 0.1676 (0.2228)  loss: 0.7830 (0.7875)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [67]  [2000/6250]  eta: 0:19:11  lr: 0.000033  grad: 0.1715 (0.2215)  loss: 0.7861 (0.7874)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [67]  [2100/6250]  eta: 0:18:44  lr: 0.000033  grad: 0.1538 (0.2220)  loss: 0.7934 (0.7875)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [67]  [2200/6250]  eta: 0:18:16  lr: 0.000033  grad: 0.1879 (0.2227)  loss: 0.7861 (0.7874)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [67]  [2300/6250]  eta: 0:17:48  lr: 0.000033  grad: 0.1889 (0.2235)  loss: 0.7817 (0.7873)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [67]  [2400/6250]  eta: 0:17:21  lr: 0.000033  grad: 0.1572 (0.2250)  loss: 0.7889 (0.7874)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [67]  [2500/6250]  eta: 0:16:54  lr: 0.000033  grad: 0.1601 (0.2258)  loss: 0.7871 (0.7874)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [67]  [2600/6250]  eta: 0:16:27  lr: 0.000033  grad: 0.1573 (0.2265)  loss: 0.7840 (0.7874)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [67]  [2700/6250]  eta: 0:15:59  lr: 0.000033  grad: 0.1654 (0.2281)  loss: 0.7890 (0.7874)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [67]  [2800/6250]  eta: 0:15:32  lr: 0.000033  grad: 0.1563 (0.2286)  loss: 0.7833 (0.7872)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [67]  [2900/6250]  eta: 0:15:05  lr: 0.000033  grad: 0.1863 (0.2296)  loss: 0.7826 (0.7871)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [67]  [3000/6250]  eta: 0:14:38  lr: 0.000033  grad: 0.1796 (0.2296)  loss: 0.7882 (0.7870)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [67]  [3100/6250]  eta: 0:14:10  lr: 0.000033  grad: 0.1608 (0.2286)  loss: 0.7874 (0.7869)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [67]  [3200/6250]  eta: 0:13:43  lr: 0.000033  grad: 0.1631 (0.2284)  loss: 0.7828 (0.7868)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [67]  [3300/6250]  eta: 0:13:16  lr: 0.000033  grad: 0.1709 (0.2283)  loss: 0.7714 (0.7866)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [67]  [3400/6250]  eta: 0:12:49  lr: 0.000033  grad: 0.1856 (0.2291)  loss: 0.7747 (0.7864)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [67]  [3500/6250]  eta: 0:12:22  lr: 0.000033  grad: 0.1843 (0.2285)  loss: 0.7792 (0.7863)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [67]  [3600/6250]  eta: 0:11:54  lr: 0.000033  grad: 0.1708 (0.2286)  loss: 0.7870 (0.7862)  time: 0.2692  data: 0.0001  max mem: 26157
Train: [67]  [3700/6250]  eta: 0:11:27  lr: 0.000033  grad: 0.1903 (0.2315)  loss: 0.7772 (0.7859)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [67]  [3800/6250]  eta: 0:11:00  lr: 0.000033  grad: 0.1855 (0.2325)  loss: 0.7809 (0.7858)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [67]  [3900/6250]  eta: 0:10:33  lr: 0.000033  grad: 0.1733 (0.2325)  loss: 0.7750 (0.7857)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [67]  [4000/6250]  eta: 0:10:07  lr: 0.000032  grad: 0.1809 (0.2330)  loss: 0.7749 (0.7856)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [67]  [4100/6250]  eta: 0:09:40  lr: 0.000032  grad: 0.1727 (0.2344)  loss: 0.7816 (0.7855)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [67]  [4200/6250]  eta: 0:09:13  lr: 0.000032  grad: 0.1649 (0.2336)  loss: 0.7707 (0.7853)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [67]  [4300/6250]  eta: 0:08:46  lr: 0.000032  grad: 0.1947 (0.2352)  loss: 0.7801 (0.7852)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [67]  [4400/6250]  eta: 0:08:19  lr: 0.000032  grad: 0.1852 (0.2358)  loss: 0.7900 (0.7851)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [67]  [4500/6250]  eta: 0:07:52  lr: 0.000032  grad: 0.1790 (0.2367)  loss: 0.7780 (0.7850)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [67]  [4600/6250]  eta: 0:07:25  lr: 0.000032  grad: 0.1746 (0.2367)  loss: 0.7818 (0.7849)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [67]  [4700/6250]  eta: 0:06:58  lr: 0.000032  grad: 0.1691 (0.2367)  loss: 0.7804 (0.7848)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [67]  [4800/6250]  eta: 0:06:31  lr: 0.000032  grad: 0.1679 (0.2367)  loss: 0.7702 (0.7847)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [67]  [4900/6250]  eta: 0:06:04  lr: 0.000032  grad: 0.1596 (0.2365)  loss: 0.7810 (0.7845)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [67]  [5000/6250]  eta: 0:05:37  lr: 0.000032  grad: 0.1760 (0.2360)  loss: 0.7715 (0.7844)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [67]  [5100/6250]  eta: 0:05:10  lr: 0.000032  grad: 0.2012 (0.2364)  loss: 0.7669 (0.7842)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [67]  [5200/6250]  eta: 0:04:43  lr: 0.000032  grad: 0.1872 (0.2367)  loss: 0.7775 (0.7840)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [67]  [5300/6250]  eta: 0:04:16  lr: 0.000032  grad: 0.1687 (0.2383)  loss: 0.7801 (0.7839)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [67]  [5400/6250]  eta: 0:03:49  lr: 0.000032  grad: 0.1844 (0.2391)  loss: 0.7750 (0.7838)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [67]  [5500/6250]  eta: 0:03:22  lr: 0.000032  grad: 0.1780 (0.2390)  loss: 0.7810 (0.7837)  time: 0.2692  data: 0.0001  max mem: 26157
Train: [67]  [5600/6250]  eta: 0:02:55  lr: 0.000032  grad: 0.1748 (0.2387)  loss: 0.7815 (0.7836)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [67]  [5700/6250]  eta: 0:02:28  lr: 0.000032  grad: 0.1701 (0.2384)  loss: 0.7750 (0.7836)  time: 0.2701  data: 0.0002  max mem: 26157
Train: [67]  [5800/6250]  eta: 0:02:01  lr: 0.000032  grad: 0.1779 (0.2384)  loss: 0.7802 (0.7835)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [67]  [5900/6250]  eta: 0:01:34  lr: 0.000032  grad: 0.1780 (0.2401)  loss: 0.7750 (0.7834)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [67]  [6000/6250]  eta: 0:01:07  lr: 0.000032  grad: 0.1647 (0.2415)  loss: 0.7880 (0.7834)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [67]  [6100/6250]  eta: 0:00:40  lr: 0.000032  grad: 0.1843 (0.2417)  loss: 0.7789 (0.7833)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [67]  [6200/6250]  eta: 0:00:13  lr: 0.000032  grad: 0.1647 (0.2414)  loss: 0.7793 (0.7832)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [67]  [6249/6250]  eta: 0:00:00  lr: 0.000032  grad: 0.1602 (0.2412)  loss: 0.7773 (0.7832)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [67] Total time: 0:28:09 (0.2704 s / it)
Averaged stats: lr: 0.000032  grad: 0.1602 (0.2412)  loss: 0.7773 (0.7832)
Eval (hcp-train-subset): [67]  [ 0/62]  eta: 0:04:36  loss: 0.8169 (0.8169)  time: 4.4562  data: 4.3707  max mem: 26157
Eval (hcp-train-subset): [67]  [61/62]  eta: 0:00:00  loss: 0.8100 (0.8095)  time: 0.1173  data: 0.0347  max mem: 26157
Eval (hcp-train-subset): [67] Total time: 0:00:12 (0.1993 s / it)
Averaged stats (hcp-train-subset): loss: 0.8100 (0.8095)
Making plots (hcp-train-subset): example=10
Eval (hcp-val): [67]  [ 0/62]  eta: 0:05:21  loss: 0.8176 (0.8176)  time: 5.1829  data: 5.0960  max mem: 26157
Eval (hcp-val): [67]  [61/62]  eta: 0:00:00  loss: 0.8232 (0.8244)  time: 0.1272  data: 0.0444  max mem: 26157
Eval (hcp-val): [67] Total time: 0:00:12 (0.2064 s / it)
Averaged stats (hcp-val): loss: 0.8232 (0.8244)
Making plots (hcp-val): example=7
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [68]  [   0/6250]  eta: 10:55:54  lr: 0.000032  grad: 0.3396 (0.3396)  loss: 0.7598 (0.7598)  time: 6.2967  data: 6.0246  max mem: 26157
Train: [68]  [ 100/6250]  eta: 0:33:54  lr: 0.000032  grad: 0.1740 (0.2244)  loss: 0.7901 (0.8028)  time: 0.2715  data: 0.0002  max mem: 26157
Train: [68]  [ 200/6250]  eta: 0:30:13  lr: 0.000032  grad: 0.1676 (0.2262)  loss: 0.7907 (0.7965)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [68]  [ 300/6250]  eta: 0:28:40  lr: 0.000032  grad: 0.1779 (0.2285)  loss: 0.7950 (0.7942)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [68]  [ 400/6250]  eta: 0:27:40  lr: 0.000032  grad: 0.1677 (0.2209)  loss: 0.7851 (0.7934)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [68]  [ 500/6250]  eta: 0:26:54  lr: 0.000032  grad: 0.1562 (0.2138)  loss: 0.7928 (0.7933)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [68]  [ 600/6250]  eta: 0:26:14  lr: 0.000032  grad: 0.1563 (0.2124)  loss: 0.8043 (0.7930)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [68]  [ 700/6250]  eta: 0:25:38  lr: 0.000032  grad: 0.1539 (0.2193)  loss: 0.7971 (0.7933)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [68]  [ 800/6250]  eta: 0:25:05  lr: 0.000032  grad: 0.1532 (0.2201)  loss: 0.7984 (0.7936)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [68]  [ 900/6250]  eta: 0:24:33  lr: 0.000032  grad: 0.1540 (0.2142)  loss: 0.7936 (0.7938)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [68]  [1000/6250]  eta: 0:24:02  lr: 0.000032  grad: 0.1775 (0.2137)  loss: 0.7895 (0.7938)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [68]  [1100/6250]  eta: 0:23:32  lr: 0.000032  grad: 0.1550 (0.2130)  loss: 0.7910 (0.7936)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [68]  [1200/6250]  eta: 0:23:02  lr: 0.000032  grad: 0.1657 (0.2152)  loss: 0.7890 (0.7934)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [68]  [1300/6250]  eta: 0:22:33  lr: 0.000031  grad: 0.1608 (0.2150)  loss: 0.7949 (0.7933)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [68]  [1400/6250]  eta: 0:22:04  lr: 0.000031  grad: 0.1929 (0.2176)  loss: 0.7877 (0.7930)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [68]  [1500/6250]  eta: 0:21:35  lr: 0.000031  grad: 0.1687 (0.2162)  loss: 0.7842 (0.7927)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [68]  [1600/6250]  eta: 0:21:07  lr: 0.000031  grad: 0.1738 (0.2196)  loss: 0.7823 (0.7923)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [68]  [1700/6250]  eta: 0:20:39  lr: 0.000031  grad: 0.1598 (0.2199)  loss: 0.7841 (0.7920)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [68]  [1800/6250]  eta: 0:20:10  lr: 0.000031  grad: 0.1711 (0.2204)  loss: 0.7825 (0.7916)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [68]  [1900/6250]  eta: 0:19:42  lr: 0.000031  grad: 0.1605 (0.2224)  loss: 0.7909 (0.7914)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [68]  [2000/6250]  eta: 0:19:14  lr: 0.000031  grad: 0.1778 (0.2236)  loss: 0.7933 (0.7911)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [68]  [2100/6250]  eta: 0:18:46  lr: 0.000031  grad: 0.1859 (0.2281)  loss: 0.7769 (0.7908)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [68]  [2200/6250]  eta: 0:18:19  lr: 0.000031  grad: 0.1608 (0.2281)  loss: 0.7877 (0.7906)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [68]  [2300/6250]  eta: 0:17:51  lr: 0.000031  grad: 0.1608 (0.2295)  loss: 0.7881 (0.7903)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [68]  [2400/6250]  eta: 0:17:23  lr: 0.000031  grad: 0.1672 (0.2298)  loss: 0.7862 (0.7900)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [68]  [2500/6250]  eta: 0:16:56  lr: 0.000031  grad: 0.1654 (0.2281)  loss: 0.7834 (0.7897)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [68]  [2600/6250]  eta: 0:16:28  lr: 0.000031  grad: 0.1778 (0.2296)  loss: 0.7807 (0.7893)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [68]  [2700/6250]  eta: 0:16:01  lr: 0.000031  grad: 0.1788 (0.2300)  loss: 0.7851 (0.7891)  time: 0.2697  data: 0.0002  max mem: 26157
Train: [68]  [2800/6250]  eta: 0:15:34  lr: 0.000031  grad: 0.1684 (0.2323)  loss: 0.7819 (0.7889)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [68]  [2900/6250]  eta: 0:15:06  lr: 0.000031  grad: 0.1927 (0.2336)  loss: 0.7806 (0.7886)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [68]  [3000/6250]  eta: 0:14:39  lr: 0.000031  grad: 0.1673 (0.2340)  loss: 0.7808 (0.7882)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [68]  [3100/6250]  eta: 0:14:12  lr: 0.000031  grad: 0.1759 (0.2351)  loss: 0.7783 (0.7877)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [68]  [3200/6250]  eta: 0:13:44  lr: 0.000031  grad: 0.1767 (0.2353)  loss: 0.7729 (0.7874)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [68]  [3300/6250]  eta: 0:13:17  lr: 0.000031  grad: 0.1703 (0.2354)  loss: 0.7835 (0.7873)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [68]  [3400/6250]  eta: 0:12:50  lr: 0.000031  grad: 0.1768 (0.2351)  loss: 0.7740 (0.7871)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [68]  [3500/6250]  eta: 0:12:23  lr: 0.000031  grad: 0.1598 (0.2353)  loss: 0.7850 (0.7869)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [68]  [3600/6250]  eta: 0:11:56  lr: 0.000031  grad: 0.1780 (0.2357)  loss: 0.7737 (0.7868)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [68]  [3700/6250]  eta: 0:11:29  lr: 0.000031  grad: 0.1755 (0.2364)  loss: 0.7902 (0.7867)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [68]  [3800/6250]  eta: 0:11:01  lr: 0.000031  grad: 0.1661 (0.2366)  loss: 0.7823 (0.7865)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [68]  [3900/6250]  eta: 0:10:34  lr: 0.000031  grad: 0.1742 (0.2385)  loss: 0.7809 (0.7863)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [68]  [4000/6250]  eta: 0:10:07  lr: 0.000031  grad: 0.1744 (0.2376)  loss: 0.7783 (0.7862)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [68]  [4100/6250]  eta: 0:09:40  lr: 0.000031  grad: 0.1915 (0.2374)  loss: 0.7834 (0.7861)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [68]  [4200/6250]  eta: 0:09:13  lr: 0.000031  grad: 0.1873 (0.2367)  loss: 0.7755 (0.7859)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [68]  [4300/6250]  eta: 0:08:46  lr: 0.000031  grad: 0.1615 (0.2359)  loss: 0.7805 (0.7858)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [68]  [4400/6250]  eta: 0:08:19  lr: 0.000031  grad: 0.1661 (0.2363)  loss: 0.7837 (0.7857)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [68]  [4500/6250]  eta: 0:07:52  lr: 0.000031  grad: 0.1700 (0.2365)  loss: 0.7779 (0.7856)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [68]  [4600/6250]  eta: 0:07:25  lr: 0.000031  grad: 0.1620 (0.2364)  loss: 0.7714 (0.7854)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [68]  [4700/6250]  eta: 0:06:58  lr: 0.000031  grad: 0.1746 (0.2366)  loss: 0.7758 (0.7853)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [68]  [4800/6250]  eta: 0:06:31  lr: 0.000030  grad: 0.1773 (0.2364)  loss: 0.7715 (0.7851)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [68]  [4900/6250]  eta: 0:06:04  lr: 0.000030  grad: 0.1727 (0.2365)  loss: 0.7784 (0.7850)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [68]  [5000/6250]  eta: 0:05:37  lr: 0.000030  grad: 0.1814 (0.2374)  loss: 0.7836 (0.7849)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [68]  [5100/6250]  eta: 0:05:10  lr: 0.000030  grad: 0.1744 (0.2394)  loss: 0.7730 (0.7847)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [68]  [5200/6250]  eta: 0:04:43  lr: 0.000030  grad: 0.1760 (0.2399)  loss: 0.7661 (0.7846)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [68]  [5300/6250]  eta: 0:04:16  lr: 0.000030  grad: 0.1651 (0.2405)  loss: 0.7751 (0.7844)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [68]  [5400/6250]  eta: 0:03:49  lr: 0.000030  grad: 0.1702 (0.2406)  loss: 0.7738 (0.7842)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [68]  [5500/6250]  eta: 0:03:22  lr: 0.000030  grad: 0.1674 (0.2401)  loss: 0.7754 (0.7841)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [68]  [5600/6250]  eta: 0:02:55  lr: 0.000030  grad: 0.1661 (0.2418)  loss: 0.7723 (0.7839)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [68]  [5700/6250]  eta: 0:02:28  lr: 0.000030  grad: 0.1716 (0.2424)  loss: 0.7792 (0.7838)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [68]  [5800/6250]  eta: 0:02:01  lr: 0.000030  grad: 0.1787 (0.2429)  loss: 0.7841 (0.7837)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [68]  [5900/6250]  eta: 0:01:34  lr: 0.000030  grad: 0.1677 (0.2432)  loss: 0.7796 (0.7836)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [68]  [6000/6250]  eta: 0:01:07  lr: 0.000030  grad: 0.1698 (0.2425)  loss: 0.7814 (0.7835)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [68]  [6100/6250]  eta: 0:00:40  lr: 0.000030  grad: 0.1653 (0.2422)  loss: 0.7811 (0.7835)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [68]  [6200/6250]  eta: 0:00:13  lr: 0.000030  grad: 0.1660 (0.2421)  loss: 0.7816 (0.7834)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [68]  [6249/6250]  eta: 0:00:00  lr: 0.000030  grad: 0.1685 (0.2423)  loss: 0.7786 (0.7834)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [68] Total time: 0:28:09 (0.2704 s / it)
Averaged stats: lr: 0.000030  grad: 0.1685 (0.2423)  loss: 0.7786 (0.7834)
Eval (hcp-train-subset): [68]  [ 0/62]  eta: 0:05:07  loss: 0.8197 (0.8197)  time: 4.9578  data: 4.8745  max mem: 26157
Eval (hcp-train-subset): [68]  [61/62]  eta: 0:00:00  loss: 0.8043 (0.8069)  time: 0.0879  data: 0.0055  max mem: 26157
Eval (hcp-train-subset): [68] Total time: 0:00:11 (0.1821 s / it)
Averaged stats (hcp-train-subset): loss: 0.8043 (0.8069)
Making plots (hcp-train-subset): example=30
Eval (hcp-val): [68]  [ 0/62]  eta: 0:03:19  loss: 0.8240 (0.8240)  time: 3.2128  data: 3.1080  max mem: 26157
Eval (hcp-val): [68]  [61/62]  eta: 0:00:00  loss: 0.8231 (0.8246)  time: 0.1029  data: 0.0188  max mem: 26157
Eval (hcp-val): [68] Total time: 0:00:11 (0.1899 s / it)
Averaged stats (hcp-val): loss: 0.8231 (0.8246)
Making plots (hcp-val): example=56
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [69]  [   0/6250]  eta: 9:44:32  lr: 0.000030  grad: 0.2337 (0.2337)  loss: 0.8156 (0.8156)  time: 5.6117  data: 5.3311  max mem: 26157
Train: [69]  [ 100/6250]  eta: 0:33:01  lr: 0.000030  grad: 0.1578 (0.2654)  loss: 0.8021 (0.8021)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [69]  [ 200/6250]  eta: 0:29:49  lr: 0.000030  grad: 0.1773 (0.2423)  loss: 0.7876 (0.7974)  time: 0.2697  data: 0.0001  max mem: 26157
Train: [69]  [ 300/6250]  eta: 0:28:24  lr: 0.000030  grad: 0.1868 (0.2464)  loss: 0.7803 (0.7929)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [69]  [ 400/6250]  eta: 0:27:28  lr: 0.000030  grad: 0.1988 (0.2562)  loss: 0.7794 (0.7894)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [69]  [ 500/6250]  eta: 0:26:44  lr: 0.000030  grad: 0.1851 (0.2548)  loss: 0.7796 (0.7878)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [69]  [ 600/6250]  eta: 0:26:06  lr: 0.000030  grad: 0.1715 (0.2624)  loss: 0.7860 (0.7864)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [69]  [ 700/6250]  eta: 0:25:31  lr: 0.000030  grad: 0.1819 (0.2740)  loss: 0.7772 (0.7854)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [69]  [ 800/6250]  eta: 0:24:58  lr: 0.000030  grad: 0.1837 (0.2772)  loss: 0.7815 (0.7850)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [69]  [ 900/6250]  eta: 0:24:26  lr: 0.000030  grad: 0.1633 (0.2706)  loss: 0.7815 (0.7842)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [69]  [1000/6250]  eta: 0:23:55  lr: 0.000030  grad: 0.1852 (0.2701)  loss: 0.7786 (0.7838)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [69]  [1100/6250]  eta: 0:23:25  lr: 0.000030  grad: 0.1646 (0.2640)  loss: 0.7770 (0.7832)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [69]  [1200/6250]  eta: 0:22:56  lr: 0.000030  grad: 0.1701 (0.2651)  loss: 0.7824 (0.7825)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [69]  [1300/6250]  eta: 0:22:26  lr: 0.000030  grad: 0.1714 (0.2650)  loss: 0.7803 (0.7823)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [69]  [1400/6250]  eta: 0:21:57  lr: 0.000030  grad: 0.1639 (0.2613)  loss: 0.7729 (0.7818)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [69]  [1500/6250]  eta: 0:21:29  lr: 0.000030  grad: 0.1663 (0.2602)  loss: 0.7896 (0.7815)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [69]  [1600/6250]  eta: 0:21:01  lr: 0.000030  grad: 0.1600 (0.2648)  loss: 0.7726 (0.7810)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [69]  [1700/6250]  eta: 0:20:33  lr: 0.000030  grad: 0.1891 (0.2655)  loss: 0.7729 (0.7808)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [69]  [1800/6250]  eta: 0:20:05  lr: 0.000030  grad: 0.1689 (0.2654)  loss: 0.7811 (0.7805)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [69]  [1900/6250]  eta: 0:19:37  lr: 0.000030  grad: 0.1803 (0.2693)  loss: 0.7791 (0.7804)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [69]  [2000/6250]  eta: 0:19:09  lr: 0.000030  grad: 0.1782 (0.2714)  loss: 0.7813 (0.7803)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [69]  [2100/6250]  eta: 0:18:42  lr: 0.000029  grad: 0.1630 (0.2698)  loss: 0.7811 (0.7803)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [69]  [2200/6250]  eta: 0:18:14  lr: 0.000029  grad: 0.1862 (0.2696)  loss: 0.7865 (0.7803)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [69]  [2300/6250]  eta: 0:17:47  lr: 0.000029  grad: 0.1701 (0.2689)  loss: 0.7787 (0.7803)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [69]  [2400/6250]  eta: 0:17:19  lr: 0.000029  grad: 0.1722 (0.2683)  loss: 0.7822 (0.7803)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [69]  [2500/6250]  eta: 0:16:52  lr: 0.000029  grad: 0.1637 (0.2685)  loss: 0.7916 (0.7805)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [69]  [2600/6250]  eta: 0:16:25  lr: 0.000029  grad: 0.1604 (0.2677)  loss: 0.7879 (0.7806)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [69]  [2700/6250]  eta: 0:15:57  lr: 0.000029  grad: 0.1774 (0.2669)  loss: 0.7819 (0.7808)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [69]  [2800/6250]  eta: 0:15:30  lr: 0.000029  grad: 0.1848 (0.2657)  loss: 0.7805 (0.7809)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [69]  [2900/6250]  eta: 0:15:03  lr: 0.000029  grad: 0.1607 (0.2663)  loss: 0.7905 (0.7811)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [69]  [3000/6250]  eta: 0:14:36  lr: 0.000029  grad: 0.1685 (0.2660)  loss: 0.7839 (0.7812)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [69]  [3100/6250]  eta: 0:14:08  lr: 0.000029  grad: 0.1713 (0.2646)  loss: 0.7886 (0.7813)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [69]  [3200/6250]  eta: 0:13:41  lr: 0.000029  grad: 0.1740 (0.2631)  loss: 0.7828 (0.7814)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [69]  [3300/6250]  eta: 0:13:14  lr: 0.000029  grad: 0.1909 (0.2648)  loss: 0.7778 (0.7814)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [69]  [3400/6250]  eta: 0:12:47  lr: 0.000029  grad: 0.1763 (0.2643)  loss: 0.7803 (0.7815)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [69]  [3500/6250]  eta: 0:12:20  lr: 0.000029  grad: 0.1637 (0.2633)  loss: 0.7826 (0.7815)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [69]  [3600/6250]  eta: 0:11:53  lr: 0.000029  grad: 0.1744 (0.2636)  loss: 0.7849 (0.7816)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [69]  [3700/6250]  eta: 0:11:26  lr: 0.000029  grad: 0.1703 (0.2624)  loss: 0.7853 (0.7817)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [69]  [3800/6250]  eta: 0:10:59  lr: 0.000029  grad: 0.1719 (0.2641)  loss: 0.7780 (0.7817)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [69]  [3900/6250]  eta: 0:10:32  lr: 0.000029  grad: 0.1754 (0.2653)  loss: 0.7763 (0.7817)  time: 0.2694  data: 0.0002  max mem: 26157
Train: [69]  [4000/6250]  eta: 0:10:05  lr: 0.000029  grad: 0.1747 (0.2659)  loss: 0.7825 (0.7817)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [69]  [4100/6250]  eta: 0:09:38  lr: 0.000029  grad: 0.1830 (0.2674)  loss: 0.7850 (0.7817)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [69]  [4200/6250]  eta: 0:09:11  lr: 0.000029  grad: 0.1673 (0.2669)  loss: 0.7824 (0.7816)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [69]  [4300/6250]  eta: 0:08:44  lr: 0.000029  grad: 0.1796 (0.2677)  loss: 0.7792 (0.7816)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [69]  [4400/6250]  eta: 0:08:17  lr: 0.000029  grad: 0.1664 (0.2678)  loss: 0.7809 (0.7816)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [69]  [4500/6250]  eta: 0:07:50  lr: 0.000029  grad: 0.1882 (0.2689)  loss: 0.7841 (0.7816)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [69]  [4600/6250]  eta: 0:07:23  lr: 0.000029  grad: 0.1792 (0.2682)  loss: 0.7821 (0.7815)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [69]  [4700/6250]  eta: 0:06:56  lr: 0.000029  grad: 0.1866 (0.2680)  loss: 0.7699 (0.7815)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [69]  [4800/6250]  eta: 0:06:29  lr: 0.000029  grad: 0.1803 (0.2685)  loss: 0.7813 (0.7814)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [69]  [4900/6250]  eta: 0:06:02  lr: 0.000029  grad: 0.1817 (0.2684)  loss: 0.7700 (0.7813)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [69]  [5000/6250]  eta: 0:05:36  lr: 0.000029  grad: 0.1778 (0.2696)  loss: 0.7776 (0.7812)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [69]  [5100/6250]  eta: 0:05:09  lr: 0.000029  grad: 0.1804 (0.2699)  loss: 0.7778 (0.7812)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [69]  [5200/6250]  eta: 0:04:42  lr: 0.000029  grad: 0.2092 (0.2703)  loss: 0.7813 (0.7811)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [69]  [5300/6250]  eta: 0:04:15  lr: 0.000029  grad: 0.1933 (0.2711)  loss: 0.7798 (0.7810)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [69]  [5400/6250]  eta: 0:03:48  lr: 0.000029  grad: 0.1750 (0.2711)  loss: 0.7843 (0.7810)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [69]  [5500/6250]  eta: 0:03:21  lr: 0.000029  grad: 0.1775 (0.2712)  loss: 0.7713 (0.7809)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [69]  [5600/6250]  eta: 0:02:54  lr: 0.000028  grad: 0.1849 (0.2710)  loss: 0.7782 (0.7808)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [69]  [5700/6250]  eta: 0:02:27  lr: 0.000028  grad: 0.1737 (0.2708)  loss: 0.7773 (0.7808)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [69]  [5800/6250]  eta: 0:02:00  lr: 0.000028  grad: 0.1852 (0.2706)  loss: 0.7831 (0.7807)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [69]  [5900/6250]  eta: 0:01:34  lr: 0.000028  grad: 0.1729 (0.2706)  loss: 0.7702 (0.7806)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [69]  [6000/6250]  eta: 0:01:07  lr: 0.000028  grad: 0.1848 (0.2700)  loss: 0.7802 (0.7806)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [69]  [6100/6250]  eta: 0:00:40  lr: 0.000028  grad: 0.1646 (0.2695)  loss: 0.7848 (0.7806)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [69]  [6200/6250]  eta: 0:00:13  lr: 0.000028  grad: 0.1748 (0.2693)  loss: 0.7841 (0.7806)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [69]  [6249/6250]  eta: 0:00:00  lr: 0.000028  grad: 0.1737 (0.2693)  loss: 0.7724 (0.7806)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [69] Total time: 0:28:02 (0.2692 s / it)
Averaged stats: lr: 0.000028  grad: 0.1737 (0.2693)  loss: 0.7724 (0.7806)
Eval (hcp-train-subset): [69]  [ 0/62]  eta: 0:04:21  loss: 0.8197 (0.8197)  time: 4.2228  data: 4.1394  max mem: 26157
Eval (hcp-train-subset): [69]  [61/62]  eta: 0:00:00  loss: 0.8066 (0.8079)  time: 0.1187  data: 0.0361  max mem: 26157
Eval (hcp-train-subset): [69] Total time: 0:00:11 (0.1873 s / it)
Averaged stats (hcp-train-subset): loss: 0.8066 (0.8079)
Making plots (hcp-train-subset): example=4
Eval (hcp-val): [69]  [ 0/62]  eta: 0:05:13  loss: 0.8201 (0.8201)  time: 5.0606  data: 4.9764  max mem: 26157
Eval (hcp-val): [69]  [61/62]  eta: 0:00:00  loss: 0.8229 (0.8243)  time: 0.1256  data: 0.0411  max mem: 26157
Eval (hcp-val): [69] Total time: 0:00:11 (0.1911 s / it)
Averaged stats (hcp-val): loss: 0.8229 (0.8243)
Making plots (hcp-val): example=45
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [70]  [   0/6250]  eta: 6:59:29  lr: 0.000028  grad: 0.6485 (0.6485)  loss: 0.8006 (0.8006)  time: 4.0271  data: 3.6870  max mem: 26157
Train: [70]  [ 100/6250]  eta: 0:32:49  lr: 0.000028  grad: 0.2276 (0.2825)  loss: 0.7975 (0.8021)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [70]  [ 200/6250]  eta: 0:29:39  lr: 0.000028  grad: 0.2444 (0.2855)  loss: 0.7643 (0.7899)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [70]  [ 300/6250]  eta: 0:28:17  lr: 0.000028  grad: 0.2015 (0.2841)  loss: 0.7881 (0.7864)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [70]  [ 400/6250]  eta: 0:27:22  lr: 0.000028  grad: 0.1973 (0.2854)  loss: 0.7850 (0.7849)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [70]  [ 500/6250]  eta: 0:26:39  lr: 0.000028  grad: 0.1744 (0.2872)  loss: 0.7839 (0.7849)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [70]  [ 600/6250]  eta: 0:26:01  lr: 0.000028  grad: 0.1692 (0.2796)  loss: 0.7886 (0.7850)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [70]  [ 700/6250]  eta: 0:25:27  lr: 0.000028  grad: 0.1783 (0.2766)  loss: 0.7943 (0.7851)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [70]  [ 800/6250]  eta: 0:24:55  lr: 0.000028  grad: 0.1742 (0.2670)  loss: 0.7859 (0.7851)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [70]  [ 900/6250]  eta: 0:24:24  lr: 0.000028  grad: 0.1664 (0.2633)  loss: 0.7886 (0.7850)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [70]  [1000/6250]  eta: 0:23:54  lr: 0.000028  grad: 0.1691 (0.2571)  loss: 0.7813 (0.7845)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [70]  [1100/6250]  eta: 0:23:24  lr: 0.000028  grad: 0.1733 (0.2581)  loss: 0.7830 (0.7841)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [70]  [1200/6250]  eta: 0:22:54  lr: 0.000028  grad: 0.1688 (0.2600)  loss: 0.7842 (0.7836)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [70]  [1300/6250]  eta: 0:22:25  lr: 0.000028  grad: 0.1732 (0.2581)  loss: 0.7753 (0.7832)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [70]  [1400/6250]  eta: 0:21:56  lr: 0.000028  grad: 0.1649 (0.2567)  loss: 0.7746 (0.7828)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [70]  [1500/6250]  eta: 0:21:28  lr: 0.000028  grad: 0.1700 (0.2555)  loss: 0.7793 (0.7824)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [70]  [1600/6250]  eta: 0:21:00  lr: 0.000028  grad: 0.1757 (0.2597)  loss: 0.7699 (0.7821)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [70]  [1700/6250]  eta: 0:20:31  lr: 0.000028  grad: 0.1798 (0.2606)  loss: 0.7687 (0.7817)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [70]  [1800/6250]  eta: 0:20:04  lr: 0.000028  grad: 0.1790 (0.2603)  loss: 0.7762 (0.7813)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [70]  [1900/6250]  eta: 0:19:36  lr: 0.000028  grad: 0.1643 (0.2589)  loss: 0.7839 (0.7811)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [70]  [2000/6250]  eta: 0:19:08  lr: 0.000028  grad: 0.1774 (0.2573)  loss: 0.7778 (0.7809)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [70]  [2100/6250]  eta: 0:18:41  lr: 0.000028  grad: 0.1813 (0.2571)  loss: 0.7793 (0.7806)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [70]  [2200/6250]  eta: 0:18:13  lr: 0.000028  grad: 0.1726 (0.2561)  loss: 0.7822 (0.7804)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [70]  [2300/6250]  eta: 0:17:46  lr: 0.000028  grad: 0.1750 (0.2584)  loss: 0.7770 (0.7804)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [70]  [2400/6250]  eta: 0:17:18  lr: 0.000028  grad: 0.1849 (0.2581)  loss: 0.7835 (0.7803)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [70]  [2500/6250]  eta: 0:16:51  lr: 0.000028  grad: 0.1728 (0.2578)  loss: 0.7804 (0.7802)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [70]  [2600/6250]  eta: 0:16:24  lr: 0.000028  grad: 0.1731 (0.2585)  loss: 0.7801 (0.7803)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [70]  [2700/6250]  eta: 0:15:56  lr: 0.000028  grad: 0.1713 (0.2596)  loss: 0.7810 (0.7804)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [70]  [2800/6250]  eta: 0:15:29  lr: 0.000028  grad: 0.1790 (0.2591)  loss: 0.7848 (0.7806)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [70]  [2900/6250]  eta: 0:15:02  lr: 0.000028  grad: 0.1717 (0.2604)  loss: 0.7814 (0.7807)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [70]  [3000/6250]  eta: 0:14:35  lr: 0.000027  grad: 0.1748 (0.2602)  loss: 0.7855 (0.7808)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [70]  [3100/6250]  eta: 0:14:08  lr: 0.000027  grad: 0.1769 (0.2596)  loss: 0.7896 (0.7808)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [70]  [3200/6250]  eta: 0:13:41  lr: 0.000027  grad: 0.1701 (0.2600)  loss: 0.7911 (0.7810)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [70]  [3300/6250]  eta: 0:13:14  lr: 0.000027  grad: 0.1817 (0.2604)  loss: 0.7868 (0.7811)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [70]  [3400/6250]  eta: 0:12:46  lr: 0.000027  grad: 0.1716 (0.2587)  loss: 0.7798 (0.7812)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [70]  [3500/6250]  eta: 0:12:20  lr: 0.000027  grad: 0.1671 (0.2603)  loss: 0.7849 (0.7813)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [70]  [3600/6250]  eta: 0:11:52  lr: 0.000027  grad: 0.1720 (0.2598)  loss: 0.7879 (0.7815)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [70]  [3700/6250]  eta: 0:11:25  lr: 0.000027  grad: 0.1900 (0.2593)  loss: 0.7797 (0.7815)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [70]  [3800/6250]  eta: 0:10:58  lr: 0.000027  grad: 0.1743 (0.2597)  loss: 0.7814 (0.7816)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [70]  [3900/6250]  eta: 0:10:31  lr: 0.000027  grad: 0.1855 (0.2608)  loss: 0.7760 (0.7816)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [70]  [4000/6250]  eta: 0:10:04  lr: 0.000027  grad: 0.1821 (0.2617)  loss: 0.7808 (0.7818)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [70]  [4100/6250]  eta: 0:09:38  lr: 0.000027  grad: 0.1740 (0.2620)  loss: 0.7851 (0.7819)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [70]  [4200/6250]  eta: 0:09:11  lr: 0.000027  grad: 0.1733 (0.2626)  loss: 0.7824 (0.7820)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [70]  [4300/6250]  eta: 0:08:44  lr: 0.000027  grad: 0.1680 (0.2616)  loss: 0.7863 (0.7820)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [70]  [4400/6250]  eta: 0:08:17  lr: 0.000027  grad: 0.1716 (0.2630)  loss: 0.7774 (0.7821)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [70]  [4500/6250]  eta: 0:07:50  lr: 0.000027  grad: 0.1884 (0.2629)  loss: 0.7838 (0.7821)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [70]  [4600/6250]  eta: 0:07:23  lr: 0.000027  grad: 0.1803 (0.2642)  loss: 0.7848 (0.7821)  time: 0.2662  data: 0.0002  max mem: 26157
Train: [70]  [4700/6250]  eta: 0:06:56  lr: 0.000027  grad: 0.2300 (0.2663)  loss: 0.7855 (0.7821)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [70]  [4800/6250]  eta: 0:06:29  lr: 0.000027  grad: 0.2003 (0.2672)  loss: 0.7834 (0.7822)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [70]  [4900/6250]  eta: 0:06:02  lr: 0.000027  grad: 0.2095 (0.2688)  loss: 0.7808 (0.7822)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [70]  [5000/6250]  eta: 0:05:35  lr: 0.000027  grad: 0.2251 (0.2696)  loss: 0.7954 (0.7823)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [70]  [5100/6250]  eta: 0:05:08  lr: 0.000027  grad: 0.2442 (0.2703)  loss: 0.7834 (0.7824)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [70]  [5200/6250]  eta: 0:04:41  lr: 0.000027  grad: 0.1858 (0.2701)  loss: 0.7912 (0.7825)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [70]  [5300/6250]  eta: 0:04:15  lr: 0.000027  grad: 0.1914 (0.2702)  loss: 0.7844 (0.7827)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [70]  [5400/6250]  eta: 0:03:48  lr: 0.000027  grad: 0.1761 (0.2703)  loss: 0.7827 (0.7827)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [70]  [5500/6250]  eta: 0:03:21  lr: 0.000027  grad: 0.1753 (0.2703)  loss: 0.7863 (0.7828)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [70]  [5600/6250]  eta: 0:02:54  lr: 0.000027  grad: 0.2347 (0.2700)  loss: 0.7807 (0.7828)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [70]  [5700/6250]  eta: 0:02:27  lr: 0.000027  grad: 0.2242 (0.2702)  loss: 0.7844 (0.7829)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [70]  [5800/6250]  eta: 0:02:00  lr: 0.000027  grad: 0.1831 (0.2699)  loss: 0.7871 (0.7829)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [70]  [5900/6250]  eta: 0:01:33  lr: 0.000027  grad: 0.1816 (0.2697)  loss: 0.7853 (0.7830)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [70]  [6000/6250]  eta: 0:01:07  lr: 0.000027  grad: 0.2050 (0.2700)  loss: 0.7836 (0.7831)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [70]  [6100/6250]  eta: 0:00:40  lr: 0.000027  grad: 0.1761 (0.2695)  loss: 0.7858 (0.7831)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [70]  [6200/6250]  eta: 0:00:13  lr: 0.000027  grad: 0.1846 (0.2689)  loss: 0.7757 (0.7831)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [70]  [6249/6250]  eta: 0:00:00  lr: 0.000027  grad: 0.1838 (0.2684)  loss: 0.7896 (0.7831)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [70] Total time: 0:27:59 (0.2688 s / it)
Averaged stats: lr: 0.000027  grad: 0.1838 (0.2684)  loss: 0.7896 (0.7831)
Eval (hcp-train-subset): [70]  [ 0/62]  eta: 0:04:18  loss: 0.8167 (0.8167)  time: 4.1708  data: 4.0881  max mem: 26157
Eval (hcp-train-subset): [70]  [61/62]  eta: 0:00:00  loss: 0.8029 (0.8069)  time: 0.0888  data: 0.0063  max mem: 26157
Eval (hcp-train-subset): [70] Total time: 0:00:10 (0.1724 s / it)
Averaged stats (hcp-train-subset): loss: 0.8029 (0.8069)
Making plots (hcp-train-subset): example=31
Eval (hcp-val): [70]  [ 0/62]  eta: 0:02:47  loss: 0.8197 (0.8197)  time: 2.6994  data: 2.5906  max mem: 26157
Eval (hcp-val): [70]  [61/62]  eta: 0:00:00  loss: 0.8244 (0.8255)  time: 0.0912  data: 0.0090  max mem: 26157
Eval (hcp-val): [70] Total time: 0:00:10 (0.1721 s / it)
Averaged stats (hcp-val): loss: 0.8244 (0.8255)
Making plots (hcp-val): example=3
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [71]  [   0/6250]  eta: 7:53:11  lr: 0.000027  grad: 0.6288 (0.6288)  loss: 0.8223 (0.8223)  time: 4.5426  data: 4.2690  max mem: 26157
Train: [71]  [ 100/6250]  eta: 0:32:08  lr: 0.000027  grad: 0.2224 (0.3828)  loss: 0.8120 (0.8058)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [71]  [ 200/6250]  eta: 0:29:17  lr: 0.000027  grad: 0.2531 (0.3519)  loss: 0.7995 (0.8008)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [71]  [ 300/6250]  eta: 0:28:03  lr: 0.000027  grad: 0.1851 (0.3170)  loss: 0.7972 (0.7980)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [71]  [ 400/6250]  eta: 0:27:12  lr: 0.000026  grad: 0.1976 (0.3080)  loss: 0.7908 (0.7963)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [71]  [ 500/6250]  eta: 0:26:30  lr: 0.000026  grad: 0.1804 (0.2902)  loss: 0.7882 (0.7954)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [71]  [ 600/6250]  eta: 0:25:54  lr: 0.000026  grad: 0.1759 (0.2934)  loss: 0.7929 (0.7948)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [ 700/6250]  eta: 0:25:20  lr: 0.000026  grad: 0.1750 (0.2886)  loss: 0.7901 (0.7938)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [71]  [ 800/6250]  eta: 0:24:49  lr: 0.000026  grad: 0.1833 (0.2787)  loss: 0.7868 (0.7932)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [71]  [ 900/6250]  eta: 0:24:18  lr: 0.000026  grad: 0.1869 (0.2721)  loss: 0.7886 (0.7925)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [71]  [1000/6250]  eta: 0:23:48  lr: 0.000026  grad: 0.2022 (0.2721)  loss: 0.7847 (0.7918)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [71]  [1100/6250]  eta: 0:23:19  lr: 0.000026  grad: 0.1825 (0.2685)  loss: 0.7767 (0.7910)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [71]  [1200/6250]  eta: 0:22:50  lr: 0.000026  grad: 0.1822 (0.2703)  loss: 0.7869 (0.7904)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [71]  [1300/6250]  eta: 0:22:21  lr: 0.000026  grad: 0.1818 (0.2697)  loss: 0.7799 (0.7898)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [1400/6250]  eta: 0:21:52  lr: 0.000026  grad: 0.1781 (0.2691)  loss: 0.7793 (0.7892)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [71]  [1500/6250]  eta: 0:21:24  lr: 0.000026  grad: 0.1735 (0.2681)  loss: 0.7854 (0.7889)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [1600/6250]  eta: 0:20:56  lr: 0.000026  grad: 0.1754 (0.2662)  loss: 0.7806 (0.7886)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [71]  [1700/6250]  eta: 0:20:28  lr: 0.000026  grad: 0.1764 (0.2628)  loss: 0.7823 (0.7882)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [71]  [1800/6250]  eta: 0:20:01  lr: 0.000026  grad: 0.1766 (0.2609)  loss: 0.7856 (0.7878)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [71]  [1900/6250]  eta: 0:19:33  lr: 0.000026  grad: 0.1709 (0.2585)  loss: 0.7852 (0.7876)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [71]  [2000/6250]  eta: 0:19:05  lr: 0.000026  grad: 0.1679 (0.2586)  loss: 0.7749 (0.7874)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [2100/6250]  eta: 0:18:38  lr: 0.000026  grad: 0.1664 (0.2564)  loss: 0.7829 (0.7871)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [71]  [2200/6250]  eta: 0:18:10  lr: 0.000026  grad: 0.1791 (0.2549)  loss: 0.7726 (0.7868)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [71]  [2300/6250]  eta: 0:17:43  lr: 0.000026  grad: 0.1810 (0.2542)  loss: 0.7721 (0.7865)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [71]  [2400/6250]  eta: 0:17:16  lr: 0.000026  grad: 0.1807 (0.2535)  loss: 0.7732 (0.7861)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [71]  [2500/6250]  eta: 0:16:49  lr: 0.000026  grad: 0.1711 (0.2530)  loss: 0.7813 (0.7859)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [71]  [2600/6250]  eta: 0:16:22  lr: 0.000026  grad: 0.1698 (0.2553)  loss: 0.7862 (0.7856)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [71]  [2700/6250]  eta: 0:15:55  lr: 0.000026  grad: 0.1798 (0.2552)  loss: 0.7804 (0.7853)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [71]  [2800/6250]  eta: 0:15:27  lr: 0.000026  grad: 0.1603 (0.2558)  loss: 0.7856 (0.7852)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [2900/6250]  eta: 0:15:00  lr: 0.000026  grad: 0.1747 (0.2543)  loss: 0.7821 (0.7850)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [71]  [3000/6250]  eta: 0:14:33  lr: 0.000026  grad: 0.1717 (0.2538)  loss: 0.7796 (0.7848)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [71]  [3100/6250]  eta: 0:14:06  lr: 0.000026  grad: 0.1807 (0.2535)  loss: 0.7811 (0.7847)  time: 0.2715  data: 0.0001  max mem: 26157
Train: [71]  [3200/6250]  eta: 0:13:39  lr: 0.000026  grad: 0.1684 (0.2539)  loss: 0.7825 (0.7846)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [71]  [3300/6250]  eta: 0:13:12  lr: 0.000026  grad: 0.1675 (0.2538)  loss: 0.7799 (0.7844)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [3400/6250]  eta: 0:12:45  lr: 0.000026  grad: 0.1709 (0.2549)  loss: 0.7825 (0.7844)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [71]  [3500/6250]  eta: 0:12:18  lr: 0.000026  grad: 0.1686 (0.2541)  loss: 0.7828 (0.7844)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [71]  [3600/6250]  eta: 0:11:51  lr: 0.000026  grad: 0.1696 (0.2533)  loss: 0.7863 (0.7845)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [71]  [3700/6250]  eta: 0:11:24  lr: 0.000026  grad: 0.1719 (0.2540)  loss: 0.7842 (0.7845)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [3800/6250]  eta: 0:10:57  lr: 0.000026  grad: 0.1828 (0.2539)  loss: 0.7849 (0.7845)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [71]  [3900/6250]  eta: 0:10:30  lr: 0.000026  grad: 0.1679 (0.2535)  loss: 0.7840 (0.7846)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [71]  [4000/6250]  eta: 0:10:03  lr: 0.000026  grad: 0.1701 (0.2527)  loss: 0.7842 (0.7847)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [4100/6250]  eta: 0:09:37  lr: 0.000026  grad: 0.2025 (0.2540)  loss: 0.7829 (0.7847)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [71]  [4200/6250]  eta: 0:09:10  lr: 0.000025  grad: 0.1743 (0.2546)  loss: 0.7918 (0.7847)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [71]  [4300/6250]  eta: 0:08:43  lr: 0.000025  grad: 0.1791 (0.2537)  loss: 0.7840 (0.7847)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [71]  [4400/6250]  eta: 0:08:16  lr: 0.000025  grad: 0.1868 (0.2531)  loss: 0.7878 (0.7847)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [71]  [4500/6250]  eta: 0:07:49  lr: 0.000025  grad: 0.1656 (0.2534)  loss: 0.7952 (0.7848)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [71]  [4600/6250]  eta: 0:07:22  lr: 0.000025  grad: 0.1745 (0.2534)  loss: 0.7781 (0.7848)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [71]  [4700/6250]  eta: 0:06:55  lr: 0.000025  grad: 0.1702 (0.2536)  loss: 0.7840 (0.7848)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [71]  [4800/6250]  eta: 0:06:28  lr: 0.000025  grad: 0.1776 (0.2532)  loss: 0.7816 (0.7848)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [71]  [4900/6250]  eta: 0:06:02  lr: 0.000025  grad: 0.1742 (0.2529)  loss: 0.7846 (0.7847)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [71]  [5000/6250]  eta: 0:05:35  lr: 0.000025  grad: 0.1765 (0.2534)  loss: 0.7793 (0.7847)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [71]  [5100/6250]  eta: 0:05:08  lr: 0.000025  grad: 0.2049 (0.2535)  loss: 0.7879 (0.7847)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [71]  [5200/6250]  eta: 0:04:41  lr: 0.000025  grad: 0.1969 (0.2537)  loss: 0.7850 (0.7847)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [71]  [5300/6250]  eta: 0:04:14  lr: 0.000025  grad: 0.1752 (0.2536)  loss: 0.7802 (0.7846)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [71]  [5400/6250]  eta: 0:03:47  lr: 0.000025  grad: 0.1772 (0.2542)  loss: 0.7858 (0.7845)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [71]  [5500/6250]  eta: 0:03:21  lr: 0.000025  grad: 0.2000 (0.2564)  loss: 0.7813 (0.7844)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [71]  [5600/6250]  eta: 0:02:54  lr: 0.000025  grad: 0.1805 (0.2563)  loss: 0.7794 (0.7843)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [71]  [5700/6250]  eta: 0:02:27  lr: 0.000025  grad: 0.1716 (0.2573)  loss: 0.7847 (0.7843)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [71]  [5800/6250]  eta: 0:02:00  lr: 0.000025  grad: 0.1910 (0.2577)  loss: 0.7772 (0.7842)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [71]  [5900/6250]  eta: 0:01:33  lr: 0.000025  grad: 0.1921 (0.2591)  loss: 0.7782 (0.7841)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [71]  [6000/6250]  eta: 0:01:06  lr: 0.000025  grad: 0.1778 (0.2594)  loss: 0.7713 (0.7839)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [71]  [6100/6250]  eta: 0:00:40  lr: 0.000025  grad: 0.1773 (0.2594)  loss: 0.7773 (0.7838)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [71]  [6200/6250]  eta: 0:00:13  lr: 0.000025  grad: 0.1785 (0.2589)  loss: 0.7753 (0.7838)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [71]  [6249/6250]  eta: 0:00:00  lr: 0.000025  grad: 0.1794 (0.2590)  loss: 0.7902 (0.7838)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [71] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000025  grad: 0.1794 (0.2590)  loss: 0.7902 (0.7838)
Eval (hcp-train-subset): [71]  [ 0/62]  eta: 0:03:52  loss: 0.8169 (0.8169)  time: 3.7569  data: 3.6734  max mem: 26157
Eval (hcp-train-subset): [71]  [61/62]  eta: 0:00:00  loss: 0.8046 (0.8080)  time: 0.0868  data: 0.0047  max mem: 26157
Eval (hcp-train-subset): [71] Total time: 0:00:10 (0.1663 s / it)
Averaged stats (hcp-train-subset): loss: 0.8046 (0.8080)
Making plots (hcp-train-subset): example=42
Eval (hcp-val): [71]  [ 0/62]  eta: 0:03:44  loss: 0.8197 (0.8197)  time: 3.6171  data: 3.5336  max mem: 26157
Eval (hcp-val): [71]  [61/62]  eta: 0:00:00  loss: 0.8220 (0.8244)  time: 0.0913  data: 0.0090  max mem: 26157
Eval (hcp-val): [71] Total time: 0:00:10 (0.1659 s / it)
Averaged stats (hcp-val): loss: 0.8220 (0.8244)
Making plots (hcp-val): example=33
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [72]  [   0/6250]  eta: 5:58:08  lr: 0.000025  grad: 0.1538 (0.1538)  loss: 0.8289 (0.8289)  time: 3.4381  data: 3.0965  max mem: 26157
Train: [72]  [ 100/6250]  eta: 0:32:00  lr: 0.000025  grad: 0.2239 (0.3139)  loss: 0.7890 (0.7995)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [72]  [ 200/6250]  eta: 0:29:13  lr: 0.000025  grad: 0.1941 (0.2794)  loss: 0.7937 (0.7953)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [72]  [ 300/6250]  eta: 0:28:00  lr: 0.000025  grad: 0.1867 (0.2649)  loss: 0.7926 (0.7920)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [72]  [ 400/6250]  eta: 0:27:10  lr: 0.000025  grad: 0.2074 (0.2596)  loss: 0.7782 (0.7904)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [72]  [ 500/6250]  eta: 0:26:28  lr: 0.000025  grad: 0.1882 (0.2561)  loss: 0.7862 (0.7897)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [72]  [ 600/6250]  eta: 0:25:53  lr: 0.000025  grad: 0.1771 (0.2502)  loss: 0.7879 (0.7894)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [72]  [ 700/6250]  eta: 0:25:19  lr: 0.000025  grad: 0.1869 (0.2458)  loss: 0.7838 (0.7891)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [72]  [ 800/6250]  eta: 0:24:47  lr: 0.000025  grad: 0.1658 (0.2454)  loss: 0.7843 (0.7892)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [72]  [ 900/6250]  eta: 0:24:17  lr: 0.000025  grad: 0.1762 (0.2419)  loss: 0.7894 (0.7891)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [72]  [1000/6250]  eta: 0:23:48  lr: 0.000025  grad: 0.1755 (0.2372)  loss: 0.7901 (0.7892)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [72]  [1100/6250]  eta: 0:23:19  lr: 0.000025  grad: 0.1735 (0.2405)  loss: 0.7900 (0.7888)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [72]  [1200/6250]  eta: 0:22:49  lr: 0.000025  grad: 0.1709 (0.2419)  loss: 0.7842 (0.7884)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [72]  [1300/6250]  eta: 0:22:21  lr: 0.000025  grad: 0.1763 (0.2423)  loss: 0.7877 (0.7879)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [72]  [1400/6250]  eta: 0:21:52  lr: 0.000025  grad: 0.1742 (0.2446)  loss: 0.7798 (0.7875)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [72]  [1500/6250]  eta: 0:21:24  lr: 0.000025  grad: 0.1752 (0.2450)  loss: 0.7793 (0.7872)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [72]  [1600/6250]  eta: 0:20:56  lr: 0.000025  grad: 0.1732 (0.2455)  loss: 0.7812 (0.7868)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [72]  [1700/6250]  eta: 0:20:29  lr: 0.000024  grad: 0.1749 (0.2508)  loss: 0.7748 (0.7865)  time: 0.2692  data: 0.0002  max mem: 26157
Train: [72]  [1800/6250]  eta: 0:20:01  lr: 0.000024  grad: 0.1671 (0.2533)  loss: 0.7778 (0.7863)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [72]  [1900/6250]  eta: 0:19:33  lr: 0.000024  grad: 0.1864 (0.2528)  loss: 0.7829 (0.7860)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [72]  [2000/6250]  eta: 0:19:06  lr: 0.000024  grad: 0.1753 (0.2511)  loss: 0.7879 (0.7859)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [72]  [2100/6250]  eta: 0:18:38  lr: 0.000024  grad: 0.1739 (0.2520)  loss: 0.7780 (0.7857)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [72]  [2200/6250]  eta: 0:18:11  lr: 0.000024  grad: 0.1916 (0.2513)  loss: 0.7839 (0.7854)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [72]  [2300/6250]  eta: 0:17:43  lr: 0.000024  grad: 0.1774 (0.2511)  loss: 0.7777 (0.7851)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [72]  [2400/6250]  eta: 0:17:16  lr: 0.000024  grad: 0.1901 (0.2499)  loss: 0.7738 (0.7849)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [72]  [2500/6250]  eta: 0:16:49  lr: 0.000024  grad: 0.1676 (0.2496)  loss: 0.7714 (0.7846)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [72]  [2600/6250]  eta: 0:16:22  lr: 0.000024  grad: 0.1788 (0.2496)  loss: 0.7796 (0.7843)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [72]  [2700/6250]  eta: 0:15:55  lr: 0.000024  grad: 0.1787 (0.2500)  loss: 0.7741 (0.7840)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [72]  [2800/6250]  eta: 0:15:27  lr: 0.000024  grad: 0.1718 (0.2509)  loss: 0.7864 (0.7836)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [72]  [2900/6250]  eta: 0:15:00  lr: 0.000024  grad: 0.1975 (0.2518)  loss: 0.7730 (0.7832)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [72]  [3000/6250]  eta: 0:14:33  lr: 0.000024  grad: 0.1947 (0.2511)  loss: 0.7733 (0.7830)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [72]  [3100/6250]  eta: 0:14:06  lr: 0.000024  grad: 0.1790 (0.2516)  loss: 0.7723 (0.7827)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [72]  [3200/6250]  eta: 0:13:40  lr: 0.000024  grad: 0.2004 (0.2538)  loss: 0.7783 (0.7825)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [72]  [3300/6250]  eta: 0:13:13  lr: 0.000024  grad: 0.1848 (0.2527)  loss: 0.7733 (0.7823)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [72]  [3400/6250]  eta: 0:12:46  lr: 0.000024  grad: 0.1629 (0.2530)  loss: 0.7837 (0.7822)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [72]  [3500/6250]  eta: 0:12:19  lr: 0.000024  grad: 0.1866 (0.2534)  loss: 0.7774 (0.7821)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [72]  [3600/6250]  eta: 0:11:52  lr: 0.000024  grad: 0.1762 (0.2530)  loss: 0.7836 (0.7820)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [72]  [3700/6250]  eta: 0:11:25  lr: 0.000024  grad: 0.1907 (0.2528)  loss: 0.7768 (0.7819)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [72]  [3800/6250]  eta: 0:10:58  lr: 0.000024  grad: 0.1850 (0.2530)  loss: 0.7783 (0.7818)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [72]  [3900/6250]  eta: 0:10:31  lr: 0.000024  grad: 0.1769 (0.2570)  loss: 0.7782 (0.7817)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [72]  [4000/6250]  eta: 0:10:04  lr: 0.000024  grad: 0.1853 (0.2575)  loss: 0.7803 (0.7815)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [72]  [4100/6250]  eta: 0:09:37  lr: 0.000024  grad: 0.1841 (0.2575)  loss: 0.7757 (0.7814)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [72]  [4200/6250]  eta: 0:09:10  lr: 0.000024  grad: 0.1817 (0.2565)  loss: 0.7750 (0.7812)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [72]  [4300/6250]  eta: 0:08:43  lr: 0.000024  grad: 0.2171 (0.2570)  loss: 0.7769 (0.7811)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [72]  [4400/6250]  eta: 0:08:16  lr: 0.000024  grad: 0.1959 (0.2584)  loss: 0.7800 (0.7810)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [72]  [4500/6250]  eta: 0:07:50  lr: 0.000024  grad: 0.1857 (0.2585)  loss: 0.7815 (0.7809)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [72]  [4600/6250]  eta: 0:07:23  lr: 0.000024  grad: 0.1811 (0.2581)  loss: 0.7729 (0.7808)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [72]  [4700/6250]  eta: 0:06:56  lr: 0.000024  grad: 0.2137 (0.2609)  loss: 0.7853 (0.7807)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [72]  [4800/6250]  eta: 0:06:29  lr: 0.000024  grad: 0.1722 (0.2612)  loss: 0.7734 (0.7806)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [72]  [4900/6250]  eta: 0:06:02  lr: 0.000024  grad: 0.1866 (0.2630)  loss: 0.7779 (0.7805)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [72]  [5000/6250]  eta: 0:05:35  lr: 0.000024  grad: 0.1984 (0.2631)  loss: 0.7771 (0.7805)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [72]  [5100/6250]  eta: 0:05:08  lr: 0.000024  grad: 0.1833 (0.2632)  loss: 0.7663 (0.7804)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [72]  [5200/6250]  eta: 0:04:41  lr: 0.000024  grad: 0.1802 (0.2628)  loss: 0.7762 (0.7804)  time: 0.2706  data: 0.0002  max mem: 26157
Train: [72]  [5300/6250]  eta: 0:04:14  lr: 0.000024  grad: 0.1858 (0.2633)  loss: 0.7869 (0.7803)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [72]  [5400/6250]  eta: 0:03:48  lr: 0.000024  grad: 0.1817 (0.2628)  loss: 0.7811 (0.7803)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [72]  [5500/6250]  eta: 0:03:21  lr: 0.000023  grad: 0.1738 (0.2643)  loss: 0.7818 (0.7803)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [72]  [5600/6250]  eta: 0:02:54  lr: 0.000023  grad: 0.1787 (0.2637)  loss: 0.7742 (0.7803)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [72]  [5700/6250]  eta: 0:02:27  lr: 0.000023  grad: 0.1873 (0.2629)  loss: 0.7777 (0.7803)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [72]  [5800/6250]  eta: 0:02:00  lr: 0.000023  grad: 0.1686 (0.2622)  loss: 0.7847 (0.7803)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [72]  [5900/6250]  eta: 0:01:33  lr: 0.000023  grad: 0.1753 (0.2628)  loss: 0.7794 (0.7803)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [72]  [6000/6250]  eta: 0:01:07  lr: 0.000023  grad: 0.1799 (0.2629)  loss: 0.7807 (0.7803)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [72]  [6100/6250]  eta: 0:00:40  lr: 0.000023  grad: 0.1868 (0.2636)  loss: 0.7724 (0.7803)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [72]  [6200/6250]  eta: 0:00:13  lr: 0.000023  grad: 0.1910 (0.2634)  loss: 0.7728 (0.7804)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [72]  [6249/6250]  eta: 0:00:00  lr: 0.000023  grad: 0.1713 (0.2639)  loss: 0.7775 (0.7804)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [72] Total time: 0:27:59 (0.2687 s / it)
Averaged stats: lr: 0.000023  grad: 0.1713 (0.2639)  loss: 0.7775 (0.7804)
Eval (hcp-train-subset): [72]  [ 0/62]  eta: 0:02:58  loss: 0.8113 (0.8113)  time: 2.8793  data: 2.7839  max mem: 26157
Eval (hcp-train-subset): [72]  [61/62]  eta: 0:00:00  loss: 0.8032 (0.8052)  time: 0.1067  data: 0.0245  max mem: 26157
Eval (hcp-train-subset): [72] Total time: 0:00:10 (0.1694 s / it)
Averaged stats (hcp-train-subset): loss: 0.8032 (0.8052)
Making plots (hcp-train-subset): example=41
Eval (hcp-val): [72]  [ 0/62]  eta: 0:04:41  loss: 0.8223 (0.8223)  time: 4.5332  data: 4.4502  max mem: 26157
Eval (hcp-val): [72]  [61/62]  eta: 0:00:00  loss: 0.8233 (0.8239)  time: 0.0822  data: 0.0001  max mem: 26157
Eval (hcp-val): [72] Total time: 0:00:10 (0.1711 s / it)
Averaged stats (hcp-val): loss: 0.8233 (0.8239)
Making plots (hcp-val): example=38
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [73]  [   0/6250]  eta: 8:16:09  lr: 0.000023  grad: 0.1818 (0.1818)  loss: 0.8515 (0.8515)  time: 4.7631  data: 4.4778  max mem: 26157
Train: [73]  [ 100/6250]  eta: 0:32:32  lr: 0.000023  grad: 0.2206 (0.2773)  loss: 0.8116 (0.8107)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [73]  [ 200/6250]  eta: 0:29:30  lr: 0.000023  grad: 0.2015 (0.2771)  loss: 0.7890 (0.8043)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [73]  [ 300/6250]  eta: 0:28:11  lr: 0.000023  grad: 0.2019 (0.2747)  loss: 0.7881 (0.7996)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [73]  [ 400/6250]  eta: 0:27:17  lr: 0.000023  grad: 0.1994 (0.2661)  loss: 0.7830 (0.7959)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [73]  [ 500/6250]  eta: 0:26:34  lr: 0.000023  grad: 0.1788 (0.2603)  loss: 0.7907 (0.7939)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [73]  [ 600/6250]  eta: 0:25:57  lr: 0.000023  grad: 0.1923 (0.2694)  loss: 0.7888 (0.7931)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [73]  [ 700/6250]  eta: 0:25:23  lr: 0.000023  grad: 0.1777 (0.2690)  loss: 0.7855 (0.7923)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [73]  [ 800/6250]  eta: 0:24:50  lr: 0.000023  grad: 0.1735 (0.2685)  loss: 0.7889 (0.7918)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [73]  [ 900/6250]  eta: 0:24:19  lr: 0.000023  grad: 0.1917 (0.2678)  loss: 0.7872 (0.7913)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [73]  [1000/6250]  eta: 0:23:49  lr: 0.000023  grad: 0.1970 (0.2742)  loss: 0.7885 (0.7910)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [73]  [1100/6250]  eta: 0:23:19  lr: 0.000023  grad: 0.1683 (0.2704)  loss: 0.7838 (0.7907)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [73]  [1200/6250]  eta: 0:22:50  lr: 0.000023  grad: 0.1686 (0.2758)  loss: 0.7877 (0.7906)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [73]  [1300/6250]  eta: 0:22:22  lr: 0.000023  grad: 0.1839 (0.2774)  loss: 0.7815 (0.7903)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [73]  [1400/6250]  eta: 0:21:53  lr: 0.000023  grad: 0.1738 (0.2763)  loss: 0.7848 (0.7898)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [73]  [1500/6250]  eta: 0:21:25  lr: 0.000023  grad: 0.1778 (0.2792)  loss: 0.7914 (0.7896)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [73]  [1600/6250]  eta: 0:20:57  lr: 0.000023  grad: 0.1799 (0.2807)  loss: 0.7907 (0.7894)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [73]  [1700/6250]  eta: 0:20:29  lr: 0.000023  grad: 0.1793 (0.2776)  loss: 0.7826 (0.7893)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [73]  [1800/6250]  eta: 0:20:01  lr: 0.000023  grad: 0.1669 (0.2785)  loss: 0.7862 (0.7891)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [73]  [1900/6250]  eta: 0:19:34  lr: 0.000023  grad: 0.1867 (0.2763)  loss: 0.7858 (0.7890)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [73]  [2000/6250]  eta: 0:19:06  lr: 0.000023  grad: 0.1780 (0.2737)  loss: 0.7850 (0.7887)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [73]  [2100/6250]  eta: 0:18:39  lr: 0.000023  grad: 0.1921 (0.2772)  loss: 0.7842 (0.7884)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [73]  [2200/6250]  eta: 0:18:11  lr: 0.000023  grad: 0.1760 (0.2737)  loss: 0.7855 (0.7883)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [73]  [2300/6250]  eta: 0:17:44  lr: 0.000023  grad: 0.1785 (0.2733)  loss: 0.7828 (0.7881)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [73]  [2400/6250]  eta: 0:17:16  lr: 0.000023  grad: 0.1771 (0.2719)  loss: 0.7829 (0.7880)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [73]  [2500/6250]  eta: 0:16:49  lr: 0.000023  grad: 0.1706 (0.2704)  loss: 0.7828 (0.7878)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [73]  [2600/6250]  eta: 0:16:22  lr: 0.000023  grad: 0.1735 (0.2692)  loss: 0.7875 (0.7876)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [73]  [2700/6250]  eta: 0:15:55  lr: 0.000023  grad: 0.1837 (0.2710)  loss: 0.7804 (0.7875)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [73]  [2800/6250]  eta: 0:15:28  lr: 0.000023  grad: 0.1821 (0.2702)  loss: 0.7816 (0.7872)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [73]  [2900/6250]  eta: 0:15:01  lr: 0.000023  grad: 0.1746 (0.2704)  loss: 0.7803 (0.7869)  time: 0.2705  data: 0.0002  max mem: 26157
Train: [73]  [3000/6250]  eta: 0:14:34  lr: 0.000023  grad: 0.1738 (0.2771)  loss: 0.7820 (0.7868)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [73]  [3100/6250]  eta: 0:14:07  lr: 0.000023  grad: 0.1788 (0.2767)  loss: 0.7859 (0.7865)  time: 0.2691  data: 0.0002  max mem: 26157
Train: [73]  [3200/6250]  eta: 0:13:40  lr: 0.000022  grad: 0.1789 (0.2774)  loss: 0.7805 (0.7863)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [73]  [3300/6250]  eta: 0:13:12  lr: 0.000022  grad: 0.1886 (0.2776)  loss: 0.7811 (0.7861)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [73]  [3400/6250]  eta: 0:12:45  lr: 0.000022  grad: 0.1791 (0.2756)  loss: 0.7736 (0.7860)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [73]  [3500/6250]  eta: 0:12:18  lr: 0.000022  grad: 0.1887 (0.2745)  loss: 0.7780 (0.7858)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [73]  [3600/6250]  eta: 0:11:51  lr: 0.000022  grad: 0.1743 (0.2732)  loss: 0.7853 (0.7857)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [73]  [3700/6250]  eta: 0:11:24  lr: 0.000022  grad: 0.1751 (0.2734)  loss: 0.7862 (0.7856)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [73]  [3800/6250]  eta: 0:10:58  lr: 0.000022  grad: 0.1861 (0.2734)  loss: 0.7781 (0.7854)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [73]  [3900/6250]  eta: 0:10:31  lr: 0.000022  grad: 0.1764 (0.2747)  loss: 0.7836 (0.7853)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [73]  [4000/6250]  eta: 0:10:04  lr: 0.000022  grad: 0.1787 (0.2741)  loss: 0.7756 (0.7853)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [73]  [4100/6250]  eta: 0:09:37  lr: 0.000022  grad: 0.1761 (0.2764)  loss: 0.7873 (0.7853)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [73]  [4200/6250]  eta: 0:09:10  lr: 0.000022  grad: 0.1885 (0.2762)  loss: 0.7805 (0.7852)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [73]  [4300/6250]  eta: 0:08:43  lr: 0.000022  grad: 0.1910 (0.2758)  loss: 0.7823 (0.7851)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [73]  [4400/6250]  eta: 0:08:16  lr: 0.000022  grad: 0.1838 (0.2753)  loss: 0.7823 (0.7851)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [73]  [4500/6250]  eta: 0:07:49  lr: 0.000022  grad: 0.1788 (0.2757)  loss: 0.7713 (0.7850)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [73]  [4600/6250]  eta: 0:07:22  lr: 0.000022  grad: 0.1801 (0.2774)  loss: 0.7834 (0.7850)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [73]  [4700/6250]  eta: 0:06:55  lr: 0.000022  grad: 0.1795 (0.2791)  loss: 0.7718 (0.7850)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [73]  [4800/6250]  eta: 0:06:28  lr: 0.000022  grad: 0.1802 (0.2799)  loss: 0.7805 (0.7849)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [73]  [4900/6250]  eta: 0:06:02  lr: 0.000022  grad: 0.1794 (0.2796)  loss: 0.7834 (0.7847)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [73]  [5000/6250]  eta: 0:05:35  lr: 0.000022  grad: 0.1795 (0.2784)  loss: 0.7756 (0.7846)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [73]  [5100/6250]  eta: 0:05:08  lr: 0.000022  grad: 0.1961 (0.2786)  loss: 0.7757 (0.7844)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [73]  [5200/6250]  eta: 0:04:41  lr: 0.000022  grad: 0.1940 (0.2794)  loss: 0.7669 (0.7842)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [73]  [5300/6250]  eta: 0:04:14  lr: 0.000022  grad: 0.1842 (0.2796)  loss: 0.7741 (0.7840)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [73]  [5400/6250]  eta: 0:03:47  lr: 0.000022  grad: 0.1991 (0.2794)  loss: 0.7743 (0.7837)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [73]  [5500/6250]  eta: 0:03:21  lr: 0.000022  grad: 0.1965 (0.2797)  loss: 0.7713 (0.7836)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [73]  [5600/6250]  eta: 0:02:54  lr: 0.000022  grad: 0.1978 (0.2811)  loss: 0.7671 (0.7834)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [73]  [5700/6250]  eta: 0:02:27  lr: 0.000022  grad: 0.1883 (0.2808)  loss: 0.7604 (0.7832)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [73]  [5800/6250]  eta: 0:02:00  lr: 0.000022  grad: 0.1888 (0.2806)  loss: 0.7691 (0.7830)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [73]  [5900/6250]  eta: 0:01:33  lr: 0.000022  grad: 0.1933 (0.2816)  loss: 0.7718 (0.7828)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [73]  [6000/6250]  eta: 0:01:06  lr: 0.000022  grad: 0.1957 (0.2811)  loss: 0.7761 (0.7826)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [73]  [6100/6250]  eta: 0:00:40  lr: 0.000022  grad: 0.2000 (0.2811)  loss: 0.7667 (0.7825)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [73]  [6200/6250]  eta: 0:00:13  lr: 0.000022  grad: 0.2233 (0.2818)  loss: 0.7616 (0.7823)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [73]  [6249/6250]  eta: 0:00:00  lr: 0.000022  grad: 0.2907 (0.2824)  loss: 0.7661 (0.7822)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [73] Total time: 0:27:57 (0.2685 s / it)
Averaged stats: lr: 0.000022  grad: 0.2907 (0.2824)  loss: 0.7661 (0.7822)
Eval (hcp-train-subset): [73]  [ 0/62]  eta: 0:04:27  loss: 0.8159 (0.8159)  time: 4.3132  data: 4.2302  max mem: 26157
Eval (hcp-train-subset): [73]  [61/62]  eta: 0:00:00  loss: 0.8043 (0.8051)  time: 0.1010  data: 0.0189  max mem: 26157
Eval (hcp-train-subset): [73] Total time: 0:00:10 (0.1715 s / it)
Averaged stats (hcp-train-subset): loss: 0.8043 (0.8051)
Making plots (hcp-train-subset): example=4
Eval (hcp-val): [73]  [ 0/62]  eta: 0:03:25  loss: 0.8192 (0.8192)  time: 3.3115  data: 3.2027  max mem: 26157
Eval (hcp-val): [73]  [61/62]  eta: 0:00:00  loss: 0.8239 (0.8246)  time: 0.0842  data: 0.0001  max mem: 26157
Eval (hcp-val): [73] Total time: 0:00:10 (0.1659 s / it)
Averaged stats (hcp-val): loss: 0.8239 (0.8246)
Making plots (hcp-val): example=51
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [74]  [   0/6250]  eta: 7:01:28  lr: 0.000022  grad: 0.3829 (0.3829)  loss: 0.8271 (0.8271)  time: 4.0461  data: 3.7065  max mem: 26157
Train: [74]  [ 100/6250]  eta: 0:31:51  lr: 0.000022  grad: 0.2394 (0.3352)  loss: 0.7811 (0.7897)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [74]  [ 200/6250]  eta: 0:29:09  lr: 0.000022  grad: 0.2055 (0.3072)  loss: 0.7785 (0.7885)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [74]  [ 300/6250]  eta: 0:27:56  lr: 0.000022  grad: 0.2113 (0.3025)  loss: 0.7840 (0.7873)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [74]  [ 400/6250]  eta: 0:27:08  lr: 0.000022  grad: 0.2036 (0.2904)  loss: 0.7717 (0.7857)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [74]  [ 500/6250]  eta: 0:26:28  lr: 0.000022  grad: 0.1921 (0.3028)  loss: 0.7797 (0.7856)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [74]  [ 600/6250]  eta: 0:25:52  lr: 0.000022  grad: 0.1741 (0.2845)  loss: 0.7935 (0.7863)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [74]  [ 700/6250]  eta: 0:25:18  lr: 0.000022  grad: 0.1782 (0.2760)  loss: 0.7895 (0.7869)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [74]  [ 800/6250]  eta: 0:24:47  lr: 0.000022  grad: 0.1771 (0.2763)  loss: 0.7884 (0.7868)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [74]  [ 900/6250]  eta: 0:24:17  lr: 0.000021  grad: 0.1786 (0.2730)  loss: 0.7806 (0.7864)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [74]  [1000/6250]  eta: 0:23:47  lr: 0.000021  grad: 0.1810 (0.2745)  loss: 0.7758 (0.7858)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [74]  [1100/6250]  eta: 0:23:18  lr: 0.000021  grad: 0.1840 (0.2769)  loss: 0.7803 (0.7854)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [74]  [1200/6250]  eta: 0:22:49  lr: 0.000021  grad: 0.1825 (0.2739)  loss: 0.7774 (0.7850)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [74]  [1300/6250]  eta: 0:22:21  lr: 0.000021  grad: 0.1757 (0.2731)  loss: 0.7728 (0.7846)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [74]  [1400/6250]  eta: 0:21:52  lr: 0.000021  grad: 0.1883 (0.2742)  loss: 0.7774 (0.7841)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [74]  [1500/6250]  eta: 0:21:24  lr: 0.000021  grad: 0.1778 (0.2772)  loss: 0.7837 (0.7839)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [74]  [1600/6250]  eta: 0:20:56  lr: 0.000021  grad: 0.1903 (0.2775)  loss: 0.7797 (0.7836)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [1700/6250]  eta: 0:20:28  lr: 0.000021  grad: 0.1751 (0.2802)  loss: 0.7865 (0.7833)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [74]  [1800/6250]  eta: 0:20:01  lr: 0.000021  grad: 0.1832 (0.2787)  loss: 0.7664 (0.7826)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [74]  [1900/6250]  eta: 0:19:33  lr: 0.000021  grad: 0.1851 (0.2815)  loss: 0.7790 (0.7824)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [74]  [2000/6250]  eta: 0:19:06  lr: 0.000021  grad: 0.1929 (0.2824)  loss: 0.7724 (0.7820)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [74]  [2100/6250]  eta: 0:18:38  lr: 0.000021  grad: 0.1917 (0.2819)  loss: 0.7780 (0.7819)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [74]  [2200/6250]  eta: 0:18:11  lr: 0.000021  grad: 0.1686 (0.2859)  loss: 0.7867 (0.7818)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [74]  [2300/6250]  eta: 0:17:44  lr: 0.000021  grad: 0.1851 (0.2837)  loss: 0.7828 (0.7818)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [74]  [2400/6250]  eta: 0:17:16  lr: 0.000021  grad: 0.1841 (0.2826)  loss: 0.7832 (0.7819)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [2500/6250]  eta: 0:16:49  lr: 0.000021  grad: 0.1891 (0.2847)  loss: 0.7825 (0.7818)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [74]  [2600/6250]  eta: 0:16:22  lr: 0.000021  grad: 0.1725 (0.2863)  loss: 0.7786 (0.7818)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [74]  [2700/6250]  eta: 0:15:55  lr: 0.000021  grad: 0.1914 (0.2879)  loss: 0.7853 (0.7818)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [74]  [2800/6250]  eta: 0:15:28  lr: 0.000021  grad: 0.1804 (0.2862)  loss: 0.7797 (0.7818)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [2900/6250]  eta: 0:15:00  lr: 0.000021  grad: 0.1753 (0.2866)  loss: 0.7832 (0.7818)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [74]  [3000/6250]  eta: 0:14:33  lr: 0.000021  grad: 0.1891 (0.2892)  loss: 0.7833 (0.7818)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [74]  [3100/6250]  eta: 0:14:06  lr: 0.000021  grad: 0.1747 (0.2874)  loss: 0.7830 (0.7818)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [74]  [3200/6250]  eta: 0:13:39  lr: 0.000021  grad: 0.2190 (0.2917)  loss: 0.7752 (0.7817)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [3300/6250]  eta: 0:13:12  lr: 0.000021  grad: 0.1870 (0.2904)  loss: 0.7809 (0.7817)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [74]  [3400/6250]  eta: 0:12:45  lr: 0.000021  grad: 0.1856 (0.2905)  loss: 0.7879 (0.7817)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [74]  [3500/6250]  eta: 0:12:18  lr: 0.000021  grad: 0.1886 (0.2909)  loss: 0.7816 (0.7817)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [74]  [3600/6250]  eta: 0:11:51  lr: 0.000021  grad: 0.1907 (0.2924)  loss: 0.7838 (0.7817)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [74]  [3700/6250]  eta: 0:11:24  lr: 0.000021  grad: 0.1801 (0.2917)  loss: 0.7811 (0.7816)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [74]  [3800/6250]  eta: 0:10:57  lr: 0.000021  grad: 0.1917 (0.2930)  loss: 0.7821 (0.7816)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [3900/6250]  eta: 0:10:31  lr: 0.000021  grad: 0.1877 (0.2967)  loss: 0.7795 (0.7816)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [74]  [4000/6250]  eta: 0:10:04  lr: 0.000021  grad: 0.1951 (0.2983)  loss: 0.7724 (0.7815)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [74]  [4100/6250]  eta: 0:09:37  lr: 0.000021  grad: 0.1774 (0.2981)  loss: 0.7751 (0.7814)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [74]  [4200/6250]  eta: 0:09:10  lr: 0.000021  grad: 0.1908 (0.2988)  loss: 0.7773 (0.7814)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [4300/6250]  eta: 0:08:43  lr: 0.000021  grad: 0.1879 (0.2999)  loss: 0.7788 (0.7813)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [74]  [4400/6250]  eta: 0:08:16  lr: 0.000021  grad: 0.1893 (0.3031)  loss: 0.7787 (0.7812)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [74]  [4500/6250]  eta: 0:07:49  lr: 0.000021  grad: 0.1882 (0.3026)  loss: 0.7778 (0.7811)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [74]  [4600/6250]  eta: 0:07:22  lr: 0.000021  grad: 0.1887 (0.3027)  loss: 0.7805 (0.7811)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [74]  [4700/6250]  eta: 0:06:55  lr: 0.000021  grad: 0.1937 (0.3048)  loss: 0.7812 (0.7810)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [74]  [4800/6250]  eta: 0:06:29  lr: 0.000021  grad: 0.1826 (0.3040)  loss: 0.7753 (0.7810)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [74]  [4900/6250]  eta: 0:06:02  lr: 0.000020  grad: 0.1892 (0.3037)  loss: 0.7765 (0.7809)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [5000/6250]  eta: 0:05:35  lr: 0.000020  grad: 0.1939 (0.3024)  loss: 0.7731 (0.7809)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [74]  [5100/6250]  eta: 0:05:08  lr: 0.000020  grad: 0.1937 (0.3052)  loss: 0.7816 (0.7808)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [74]  [5200/6250]  eta: 0:04:41  lr: 0.000020  grad: 0.1777 (0.3087)  loss: 0.7827 (0.7809)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [74]  [5300/6250]  eta: 0:04:14  lr: 0.000020  grad: 0.1952 (0.3101)  loss: 0.7889 (0.7809)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [74]  [5400/6250]  eta: 0:03:47  lr: 0.000020  grad: 0.1843 (0.3087)  loss: 0.7799 (0.7809)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [74]  [5500/6250]  eta: 0:03:21  lr: 0.000020  grad: 0.1790 (0.3090)  loss: 0.7834 (0.7809)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [74]  [5600/6250]  eta: 0:02:54  lr: 0.000020  grad: 0.1790 (0.3094)  loss: 0.7769 (0.7808)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [74]  [5700/6250]  eta: 0:02:27  lr: 0.000020  grad: 0.1838 (0.3081)  loss: 0.7757 (0.7809)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [74]  [5800/6250]  eta: 0:02:00  lr: 0.000020  grad: 0.1819 (0.3074)  loss: 0.7818 (0.7809)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [74]  [5900/6250]  eta: 0:01:33  lr: 0.000020  grad: 0.2214 (0.3083)  loss: 0.7824 (0.7809)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [74]  [6000/6250]  eta: 0:01:07  lr: 0.000020  grad: 0.1810 (0.3072)  loss: 0.7778 (0.7809)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [74]  [6100/6250]  eta: 0:00:40  lr: 0.000020  grad: 0.1809 (0.3076)  loss: 0.7861 (0.7808)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [74]  [6200/6250]  eta: 0:00:13  lr: 0.000020  grad: 0.1862 (0.3075)  loss: 0.7782 (0.7808)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [74]  [6249/6250]  eta: 0:00:00  lr: 0.000020  grad: 0.1893 (0.3083)  loss: 0.7839 (0.7808)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [74] Total time: 0:27:58 (0.2686 s / it)
Averaged stats: lr: 0.000020  grad: 0.1893 (0.3083)  loss: 0.7839 (0.7808)
Eval (hcp-train-subset): [74]  [ 0/62]  eta: 0:04:11  loss: 0.8126 (0.8126)  time: 4.0593  data: 3.9765  max mem: 26157
Eval (hcp-train-subset): [74]  [61/62]  eta: 0:00:00  loss: 0.8005 (0.8044)  time: 0.0935  data: 0.0110  max mem: 26157
Eval (hcp-train-subset): [74] Total time: 0:00:10 (0.1652 s / it)
Averaged stats (hcp-train-subset): loss: 0.8005 (0.8044)
Making plots (hcp-train-subset): example=2
Eval (hcp-val): [74]  [ 0/62]  eta: 0:03:55  loss: 0.8210 (0.8210)  time: 3.7935  data: 3.7118  max mem: 26157
Eval (hcp-val): [74]  [61/62]  eta: 0:00:00  loss: 0.8233 (0.8242)  time: 0.0984  data: 0.0143  max mem: 26157
Eval (hcp-val): [74] Total time: 0:00:10 (0.1748 s / it)
Averaged stats (hcp-val): loss: 0.8233 (0.8242)
Making plots (hcp-val): example=50
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [75]  [   0/6250]  eta: 8:57:39  lr: 0.000020  grad: 0.1332 (0.1332)  loss: 0.8505 (0.8505)  time: 5.1616  data: 4.8855  max mem: 26157
Train: [75]  [ 100/6250]  eta: 0:32:32  lr: 0.000020  grad: 0.2018 (0.2827)  loss: 0.8025 (0.8009)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [75]  [ 200/6250]  eta: 0:29:31  lr: 0.000020  grad: 0.2051 (0.2626)  loss: 0.7908 (0.7970)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [75]  [ 300/6250]  eta: 0:28:11  lr: 0.000020  grad: 0.1901 (0.2707)  loss: 0.7926 (0.7950)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [75]  [ 400/6250]  eta: 0:27:18  lr: 0.000020  grad: 0.1896 (0.2804)  loss: 0.8007 (0.7943)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [75]  [ 500/6250]  eta: 0:26:36  lr: 0.000020  grad: 0.1946 (0.2692)  loss: 0.7982 (0.7936)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [75]  [ 600/6250]  eta: 0:25:58  lr: 0.000020  grad: 0.2144 (0.2705)  loss: 0.7951 (0.7926)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [75]  [ 700/6250]  eta: 0:25:24  lr: 0.000020  grad: 0.1804 (0.2904)  loss: 0.7861 (0.7920)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [75]  [ 800/6250]  eta: 0:24:51  lr: 0.000020  grad: 0.1875 (0.2885)  loss: 0.7856 (0.7914)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [75]  [ 900/6250]  eta: 0:24:20  lr: 0.000020  grad: 0.2083 (0.2942)  loss: 0.7876 (0.7909)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [75]  [1000/6250]  eta: 0:23:50  lr: 0.000020  grad: 0.1912 (0.2901)  loss: 0.7823 (0.7904)  time: 0.2656  data: 0.0001  max mem: 26157
Train: [75]  [1100/6250]  eta: 0:23:20  lr: 0.000020  grad: 0.1787 (0.2878)  loss: 0.7931 (0.7898)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [75]  [1200/6250]  eta: 0:22:51  lr: 0.000020  grad: 0.1769 (0.2819)  loss: 0.7817 (0.7892)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [75]  [1300/6250]  eta: 0:22:22  lr: 0.000020  grad: 0.1741 (0.2777)  loss: 0.7786 (0.7886)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [75]  [1400/6250]  eta: 0:21:53  lr: 0.000020  grad: 0.1940 (0.2759)  loss: 0.7766 (0.7882)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [75]  [1500/6250]  eta: 0:21:25  lr: 0.000020  grad: 0.1782 (0.2770)  loss: 0.7830 (0.7879)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [75]  [1600/6250]  eta: 0:20:57  lr: 0.000020  grad: 0.1864 (0.2759)  loss: 0.7818 (0.7875)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [75]  [1700/6250]  eta: 0:20:29  lr: 0.000020  grad: 0.1714 (0.2716)  loss: 0.7962 (0.7873)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [75]  [1800/6250]  eta: 0:20:01  lr: 0.000020  grad: 0.1747 (0.2707)  loss: 0.7807 (0.7871)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [75]  [1900/6250]  eta: 0:19:33  lr: 0.000020  grad: 0.1745 (0.2701)  loss: 0.7862 (0.7869)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [75]  [2000/6250]  eta: 0:19:06  lr: 0.000020  grad: 0.1767 (0.2749)  loss: 0.7809 (0.7866)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [75]  [2100/6250]  eta: 0:18:38  lr: 0.000020  grad: 0.1879 (0.2764)  loss: 0.7846 (0.7865)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [75]  [2200/6250]  eta: 0:18:11  lr: 0.000020  grad: 0.1746 (0.2747)  loss: 0.7851 (0.7863)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [75]  [2300/6250]  eta: 0:17:43  lr: 0.000020  grad: 0.1831 (0.2726)  loss: 0.7819 (0.7860)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [75]  [2400/6250]  eta: 0:17:16  lr: 0.000020  grad: 0.1953 (0.2713)  loss: 0.7769 (0.7857)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [75]  [2500/6250]  eta: 0:16:49  lr: 0.000020  grad: 0.1980 (0.2699)  loss: 0.7844 (0.7854)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [75]  [2600/6250]  eta: 0:16:22  lr: 0.000020  grad: 0.1931 (0.2713)  loss: 0.7765 (0.7851)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [75]  [2700/6250]  eta: 0:15:55  lr: 0.000020  grad: 0.1705 (0.2689)  loss: 0.7802 (0.7849)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [75]  [2800/6250]  eta: 0:15:28  lr: 0.000019  grad: 0.1825 (0.2674)  loss: 0.7755 (0.7846)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [75]  [2900/6250]  eta: 0:15:00  lr: 0.000019  grad: 0.1853 (0.2730)  loss: 0.7747 (0.7843)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [75]  [3000/6250]  eta: 0:14:33  lr: 0.000019  grad: 0.1798 (0.2731)  loss: 0.7706 (0.7841)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [75]  [3100/6250]  eta: 0:14:06  lr: 0.000019  grad: 0.1910 (0.2739)  loss: 0.7807 (0.7839)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [75]  [3200/6250]  eta: 0:13:39  lr: 0.000019  grad: 0.1864 (0.2737)  loss: 0.7742 (0.7837)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [75]  [3300/6250]  eta: 0:13:12  lr: 0.000019  grad: 0.1966 (0.2762)  loss: 0.7758 (0.7835)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [75]  [3400/6250]  eta: 0:12:45  lr: 0.000019  grad: 0.2086 (0.2775)  loss: 0.7743 (0.7834)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [75]  [3500/6250]  eta: 0:12:18  lr: 0.000019  grad: 0.1862 (0.2764)  loss: 0.7674 (0.7832)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [75]  [3600/6250]  eta: 0:11:51  lr: 0.000019  grad: 0.1882 (0.2776)  loss: 0.7799 (0.7830)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [75]  [3700/6250]  eta: 0:11:24  lr: 0.000019  grad: 0.1914 (0.2803)  loss: 0.7823 (0.7828)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [75]  [3800/6250]  eta: 0:10:57  lr: 0.000019  grad: 0.1971 (0.2819)  loss: 0.7734 (0.7827)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [75]  [3900/6250]  eta: 0:10:30  lr: 0.000019  grad: 0.1865 (0.2821)  loss: 0.7828 (0.7825)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [75]  [4000/6250]  eta: 0:10:03  lr: 0.000019  grad: 0.1902 (0.2829)  loss: 0.7787 (0.7824)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [75]  [4100/6250]  eta: 0:09:37  lr: 0.000019  grad: 0.1857 (0.2816)  loss: 0.7819 (0.7824)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [75]  [4200/6250]  eta: 0:09:10  lr: 0.000019  grad: 0.1889 (0.2834)  loss: 0.7776 (0.7824)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [75]  [4300/6250]  eta: 0:08:43  lr: 0.000019  grad: 0.1795 (0.2872)  loss: 0.7857 (0.7824)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [75]  [4400/6250]  eta: 0:08:16  lr: 0.000019  grad: 0.1732 (0.2901)  loss: 0.7931 (0.7824)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [75]  [4500/6250]  eta: 0:07:49  lr: 0.000019  grad: 0.1936 (0.2910)  loss: 0.7867 (0.7824)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [75]  [4600/6250]  eta: 0:07:22  lr: 0.000019  grad: 0.1899 (0.2934)  loss: 0.7819 (0.7824)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [75]  [4700/6250]  eta: 0:06:55  lr: 0.000019  grad: 0.1848 (0.2935)  loss: 0.7755 (0.7824)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [75]  [4800/6250]  eta: 0:06:28  lr: 0.000019  grad: 0.1985 (0.2938)  loss: 0.7834 (0.7824)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [75]  [4900/6250]  eta: 0:06:02  lr: 0.000019  grad: 0.1955 (0.2960)  loss: 0.7879 (0.7824)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [75]  [5000/6250]  eta: 0:05:35  lr: 0.000019  grad: 0.1893 (0.2976)  loss: 0.7857 (0.7824)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [75]  [5100/6250]  eta: 0:05:08  lr: 0.000019  grad: 0.1981 (0.2982)  loss: 0.7806 (0.7823)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [75]  [5200/6250]  eta: 0:04:41  lr: 0.000019  grad: 0.2295 (0.2994)  loss: 0.7793 (0.7822)  time: 0.2695  data: 0.0002  max mem: 26157
Train: [75]  [5300/6250]  eta: 0:04:14  lr: 0.000019  grad: 0.1885 (0.2985)  loss: 0.7835 (0.7822)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [75]  [5400/6250]  eta: 0:03:47  lr: 0.000019  grad: 0.1975 (0.2982)  loss: 0.7796 (0.7821)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [75]  [5500/6250]  eta: 0:03:21  lr: 0.000019  grad: 0.1956 (0.2983)  loss: 0.7821 (0.7820)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [75]  [5600/6250]  eta: 0:02:54  lr: 0.000019  grad: 0.1933 (0.2993)  loss: 0.7789 (0.7819)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [75]  [5700/6250]  eta: 0:02:27  lr: 0.000019  grad: 0.1905 (0.2986)  loss: 0.7750 (0.7818)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [75]  [5800/6250]  eta: 0:02:00  lr: 0.000019  grad: 0.1916 (0.3016)  loss: 0.7780 (0.7817)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [75]  [5900/6250]  eta: 0:01:33  lr: 0.000019  grad: 0.1911 (0.3020)  loss: 0.7793 (0.7817)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [75]  [6000/6250]  eta: 0:01:07  lr: 0.000019  grad: 0.3313 (0.3047)  loss: 0.7763 (0.7816)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [75]  [6100/6250]  eta: 0:00:40  lr: 0.000019  grad: 0.1850 (0.3048)  loss: 0.7766 (0.7816)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [75]  [6200/6250]  eta: 0:00:13  lr: 0.000019  grad: 0.1833 (0.3047)  loss: 0.7756 (0.7815)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [75]  [6249/6250]  eta: 0:00:00  lr: 0.000019  grad: 0.1865 (0.3046)  loss: 0.7758 (0.7814)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [75] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000019  grad: 0.1865 (0.3046)  loss: 0.7758 (0.7814)
Eval (hcp-train-subset): [75]  [ 0/62]  eta: 0:03:08  loss: 0.8117 (0.8117)  time: 3.0435  data: 2.9294  max mem: 26157
Eval (hcp-train-subset): [75]  [61/62]  eta: 0:00:00  loss: 0.7967 (0.8026)  time: 0.1144  data: 0.0322  max mem: 26157
Eval (hcp-train-subset): [75] Total time: 0:00:11 (0.1880 s / it)
Averaged stats (hcp-train-subset): loss: 0.7967 (0.8026)
Making plots (hcp-train-subset): example=2
Eval (hcp-val): [75]  [ 0/62]  eta: 0:05:19  loss: 0.8192 (0.8192)  time: 5.1483  data: 5.0657  max mem: 26157
Eval (hcp-val): [75]  [61/62]  eta: 0:00:00  loss: 0.8232 (0.8248)  time: 0.0824  data: 0.0001  max mem: 26157
Eval (hcp-val): [75] Total time: 0:00:11 (0.1782 s / it)
Averaged stats (hcp-val): loss: 0.8232 (0.8248)
Making plots (hcp-val): example=6
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [76]  [   0/6250]  eta: 6:40:21  lr: 0.000019  grad: 0.1271 (0.1271)  loss: 0.8582 (0.8582)  time: 3.8435  data: 3.5087  max mem: 26157
Train: [76]  [ 100/6250]  eta: 0:31:47  lr: 0.000019  grad: 0.2031 (0.2939)  loss: 0.8036 (0.8008)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [76]  [ 200/6250]  eta: 0:29:06  lr: 0.000019  grad: 0.2027 (0.2785)  loss: 0.7870 (0.7984)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [76]  [ 300/6250]  eta: 0:27:56  lr: 0.000019  grad: 0.1983 (0.3179)  loss: 0.7846 (0.7950)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [76]  [ 400/6250]  eta: 0:27:06  lr: 0.000019  grad: 0.1833 (0.2980)  loss: 0.7894 (0.7933)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [76]  [ 500/6250]  eta: 0:26:26  lr: 0.000019  grad: 0.1944 (0.2897)  loss: 0.7829 (0.7916)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [76]  [ 600/6250]  eta: 0:25:51  lr: 0.000019  grad: 0.1919 (0.2817)  loss: 0.7879 (0.7911)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [76]  [ 700/6250]  eta: 0:25:18  lr: 0.000019  grad: 0.1779 (0.2784)  loss: 0.7986 (0.7906)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [76]  [ 800/6250]  eta: 0:24:46  lr: 0.000018  grad: 0.1717 (0.2778)  loss: 0.7927 (0.7909)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [76]  [ 900/6250]  eta: 0:24:15  lr: 0.000018  grad: 0.1721 (0.2761)  loss: 0.7877 (0.7906)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [76]  [1000/6250]  eta: 0:23:46  lr: 0.000018  grad: 0.1813 (0.2916)  loss: 0.7900 (0.7904)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [76]  [1100/6250]  eta: 0:23:17  lr: 0.000018  grad: 0.1903 (0.2923)  loss: 0.7819 (0.7903)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [76]  [1200/6250]  eta: 0:22:48  lr: 0.000018  grad: 0.1792 (0.2881)  loss: 0.7862 (0.7901)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [76]  [1300/6250]  eta: 0:22:19  lr: 0.000018  grad: 0.1823 (0.2838)  loss: 0.7800 (0.7897)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [76]  [1400/6250]  eta: 0:21:51  lr: 0.000018  grad: 0.1715 (0.2888)  loss: 0.7940 (0.7893)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [76]  [1500/6250]  eta: 0:21:22  lr: 0.000018  grad: 0.1751 (0.2863)  loss: 0.7822 (0.7888)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [76]  [1600/6250]  eta: 0:20:54  lr: 0.000018  grad: 0.1715 (0.2840)  loss: 0.7828 (0.7885)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [76]  [1700/6250]  eta: 0:20:27  lr: 0.000018  grad: 0.1809 (0.2922)  loss: 0.7894 (0.7883)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [76]  [1800/6250]  eta: 0:19:59  lr: 0.000018  grad: 0.1797 (0.2883)  loss: 0.7881 (0.7880)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [76]  [1900/6250]  eta: 0:19:31  lr: 0.000018  grad: 0.1741 (0.2856)  loss: 0.7811 (0.7879)  time: 0.2663  data: 0.0002  max mem: 26157
Train: [76]  [2000/6250]  eta: 0:19:04  lr: 0.000018  grad: 0.1778 (0.2849)  loss: 0.7791 (0.7877)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [76]  [2100/6250]  eta: 0:18:36  lr: 0.000018  grad: 0.1746 (0.2829)  loss: 0.7859 (0.7875)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [76]  [2200/6250]  eta: 0:18:09  lr: 0.000018  grad: 0.1859 (0.2802)  loss: 0.7881 (0.7874)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [76]  [2300/6250]  eta: 0:17:42  lr: 0.000018  grad: 0.1766 (0.2828)  loss: 0.7821 (0.7874)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [76]  [2400/6250]  eta: 0:17:15  lr: 0.000018  grad: 0.1955 (0.2872)  loss: 0.7768 (0.7873)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [76]  [2500/6250]  eta: 0:16:47  lr: 0.000018  grad: 0.1899 (0.2863)  loss: 0.7854 (0.7871)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [76]  [2600/6250]  eta: 0:16:20  lr: 0.000018  grad: 0.1911 (0.2852)  loss: 0.7884 (0.7871)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [76]  [2700/6250]  eta: 0:15:53  lr: 0.000018  grad: 0.1818 (0.2841)  loss: 0.7806 (0.7869)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [76]  [2800/6250]  eta: 0:15:26  lr: 0.000018  grad: 0.1807 (0.2867)  loss: 0.7797 (0.7867)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [76]  [2900/6250]  eta: 0:14:59  lr: 0.000018  grad: 0.1877 (0.2885)  loss: 0.7816 (0.7865)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [76]  [3000/6250]  eta: 0:14:32  lr: 0.000018  grad: 0.1807 (0.2867)  loss: 0.7771 (0.7863)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [76]  [3100/6250]  eta: 0:14:05  lr: 0.000018  grad: 0.1818 (0.2906)  loss: 0.7745 (0.7859)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [76]  [3200/6250]  eta: 0:13:38  lr: 0.000018  grad: 0.1888 (0.2910)  loss: 0.7852 (0.7857)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [76]  [3300/6250]  eta: 0:13:11  lr: 0.000018  grad: 0.1898 (0.2925)  loss: 0.7863 (0.7855)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [76]  [3400/6250]  eta: 0:12:44  lr: 0.000018  grad: 0.2022 (0.2937)  loss: 0.7711 (0.7852)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [76]  [3500/6250]  eta: 0:12:17  lr: 0.000018  grad: 0.2094 (0.2939)  loss: 0.7832 (0.7850)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [76]  [3600/6250]  eta: 0:11:50  lr: 0.000018  grad: 0.2005 (0.2939)  loss: 0.7798 (0.7849)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [76]  [3700/6250]  eta: 0:11:24  lr: 0.000018  grad: 0.1856 (0.2944)  loss: 0.7854 (0.7848)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [76]  [3800/6250]  eta: 0:10:57  lr: 0.000018  grad: 0.1794 (0.2936)  loss: 0.7808 (0.7847)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [76]  [3900/6250]  eta: 0:10:30  lr: 0.000018  grad: 0.1825 (0.2933)  loss: 0.7788 (0.7846)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [76]  [4000/6250]  eta: 0:10:03  lr: 0.000018  grad: 0.1927 (0.2975)  loss: 0.7849 (0.7845)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [76]  [4100/6250]  eta: 0:09:36  lr: 0.000018  grad: 0.2010 (0.2988)  loss: 0.7876 (0.7845)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [76]  [4200/6250]  eta: 0:09:09  lr: 0.000018  grad: 0.1850 (0.2992)  loss: 0.7847 (0.7844)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [76]  [4300/6250]  eta: 0:08:42  lr: 0.000018  grad: 0.1867 (0.3004)  loss: 0.7822 (0.7843)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [76]  [4400/6250]  eta: 0:08:15  lr: 0.000018  grad: 0.1974 (0.3002)  loss: 0.7768 (0.7843)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [76]  [4500/6250]  eta: 0:07:49  lr: 0.000018  grad: 0.2031 (0.3012)  loss: 0.7725 (0.7842)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [76]  [4600/6250]  eta: 0:07:22  lr: 0.000018  grad: 0.1938 (0.3020)  loss: 0.7745 (0.7841)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [76]  [4700/6250]  eta: 0:06:55  lr: 0.000018  grad: 0.1860 (0.3030)  loss: 0.7740 (0.7839)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [76]  [4800/6250]  eta: 0:06:28  lr: 0.000018  grad: 0.1901 (0.3030)  loss: 0.7798 (0.7838)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [76]  [4900/6250]  eta: 0:06:01  lr: 0.000018  grad: 0.1944 (0.3039)  loss: 0.7788 (0.7837)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [76]  [5000/6250]  eta: 0:05:34  lr: 0.000018  grad: 0.1994 (0.3047)  loss: 0.7829 (0.7837)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [76]  [5100/6250]  eta: 0:05:08  lr: 0.000017  grad: 0.1940 (0.3044)  loss: 0.7812 (0.7836)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [76]  [5200/6250]  eta: 0:04:41  lr: 0.000017  grad: 0.1895 (0.3055)  loss: 0.7683 (0.7835)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [76]  [5300/6250]  eta: 0:04:14  lr: 0.000017  grad: 0.2170 (0.3094)  loss: 0.7843 (0.7834)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [76]  [5400/6250]  eta: 0:03:47  lr: 0.000017  grad: 0.2148 (0.3108)  loss: 0.7800 (0.7833)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [76]  [5500/6250]  eta: 0:03:20  lr: 0.000017  grad: 0.2098 (0.3128)  loss: 0.7768 (0.7832)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [76]  [5600/6250]  eta: 0:02:54  lr: 0.000017  grad: 0.1920 (0.3149)  loss: 0.7793 (0.7831)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [76]  [5700/6250]  eta: 0:02:27  lr: 0.000017  grad: 0.2018 (0.3185)  loss: 0.7762 (0.7831)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [76]  [5800/6250]  eta: 0:02:00  lr: 0.000017  grad: 0.2022 (0.3181)  loss: 0.7748 (0.7829)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [76]  [5900/6250]  eta: 0:01:33  lr: 0.000017  grad: 0.1875 (0.3184)  loss: 0.7762 (0.7828)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [76]  [6000/6250]  eta: 0:01:06  lr: 0.000017  grad: 0.1847 (0.3196)  loss: 0.7810 (0.7828)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [76]  [6100/6250]  eta: 0:00:40  lr: 0.000017  grad: 0.2089 (0.3196)  loss: 0.7765 (0.7827)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [76]  [6200/6250]  eta: 0:00:13  lr: 0.000017  grad: 0.1794 (0.3184)  loss: 0.7841 (0.7827)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [76]  [6249/6250]  eta: 0:00:00  lr: 0.000017  grad: 0.1854 (0.3176)  loss: 0.7766 (0.7827)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [76] Total time: 0:27:56 (0.2683 s / it)
Averaged stats: lr: 0.000017  grad: 0.1854 (0.3176)  loss: 0.7766 (0.7827)
Eval (hcp-train-subset): [76]  [ 0/62]  eta: 0:04:31  loss: 0.8060 (0.8060)  time: 4.3813  data: 4.2994  max mem: 26157
Eval (hcp-train-subset): [76]  [61/62]  eta: 0:00:00  loss: 0.7982 (0.8030)  time: 0.1110  data: 0.0287  max mem: 26157
Eval (hcp-train-subset): [76] Total time: 0:00:10 (0.1667 s / it)
Averaged stats (hcp-train-subset): loss: 0.7982 (0.8030)
Making plots (hcp-train-subset): example=35
Eval (hcp-val): [76]  [ 0/62]  eta: 0:03:22  loss: 0.8241 (0.8241)  time: 3.2631  data: 3.1684  max mem: 26157
Eval (hcp-val): [76]  [61/62]  eta: 0:00:00  loss: 0.8234 (0.8244)  time: 0.0909  data: 0.0089  max mem: 26157
Eval (hcp-val): [76] Total time: 0:00:10 (0.1670 s / it)
Averaged stats (hcp-val): loss: 0.8234 (0.8244)
Making plots (hcp-val): example=2
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [77]  [   0/6250]  eta: 7:54:06  lr: 0.000017  grad: 0.3645 (0.3645)  loss: 0.7316 (0.7316)  time: 4.5515  data: 4.2821  max mem: 26157
Train: [77]  [ 100/6250]  eta: 0:31:46  lr: 0.000017  grad: 0.2205 (0.3517)  loss: 0.8028 (0.7952)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [77]  [ 200/6250]  eta: 0:29:06  lr: 0.000017  grad: 0.1992 (0.3891)  loss: 0.7959 (0.7935)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [77]  [ 300/6250]  eta: 0:27:55  lr: 0.000017  grad: 0.1773 (0.3416)  loss: 0.7935 (0.7928)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [77]  [ 400/6250]  eta: 0:27:06  lr: 0.000017  grad: 0.1894 (0.3337)  loss: 0.7871 (0.7919)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [77]  [ 500/6250]  eta: 0:26:27  lr: 0.000017  grad: 0.1707 (0.3288)  loss: 0.7894 (0.7914)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [77]  [ 600/6250]  eta: 0:25:51  lr: 0.000017  grad: 0.1864 (0.3249)  loss: 0.7881 (0.7908)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [77]  [ 700/6250]  eta: 0:25:18  lr: 0.000017  grad: 0.1825 (0.3144)  loss: 0.7839 (0.7902)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [77]  [ 800/6250]  eta: 0:24:46  lr: 0.000017  grad: 0.1854 (0.3110)  loss: 0.7843 (0.7895)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [ 900/6250]  eta: 0:24:16  lr: 0.000017  grad: 0.1919 (0.3069)  loss: 0.7825 (0.7888)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [77]  [1000/6250]  eta: 0:23:46  lr: 0.000017  grad: 0.1899 (0.2997)  loss: 0.7781 (0.7881)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [77]  [1100/6250]  eta: 0:23:16  lr: 0.000017  grad: 0.1803 (0.2945)  loss: 0.7803 (0.7875)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [77]  [1200/6250]  eta: 0:22:47  lr: 0.000017  grad: 0.1962 (0.2979)  loss: 0.7819 (0.7868)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [77]  [1300/6250]  eta: 0:22:19  lr: 0.000017  grad: 0.1915 (0.2978)  loss: 0.7748 (0.7862)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [77]  [1400/6250]  eta: 0:21:51  lr: 0.000017  grad: 0.2003 (0.3064)  loss: 0.7847 (0.7858)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [77]  [1500/6250]  eta: 0:21:23  lr: 0.000017  grad: 0.1756 (0.3074)  loss: 0.7804 (0.7854)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [77]  [1600/6250]  eta: 0:20:55  lr: 0.000017  grad: 0.1993 (0.3028)  loss: 0.7798 (0.7851)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [77]  [1700/6250]  eta: 0:20:27  lr: 0.000017  grad: 0.1837 (0.3058)  loss: 0.7679 (0.7847)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [77]  [1800/6250]  eta: 0:20:00  lr: 0.000017  grad: 0.1878 (0.3047)  loss: 0.7817 (0.7844)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [77]  [1900/6250]  eta: 0:19:32  lr: 0.000017  grad: 0.1775 (0.3056)  loss: 0.7834 (0.7844)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [77]  [2000/6250]  eta: 0:19:05  lr: 0.000017  grad: 0.1874 (0.3027)  loss: 0.7838 (0.7843)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [2100/6250]  eta: 0:18:37  lr: 0.000017  grad: 0.1861 (0.3024)  loss: 0.7802 (0.7841)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [77]  [2200/6250]  eta: 0:18:10  lr: 0.000017  grad: 0.1771 (0.3033)  loss: 0.7828 (0.7840)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [77]  [2300/6250]  eta: 0:17:42  lr: 0.000017  grad: 0.1888 (0.3011)  loss: 0.7843 (0.7840)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [77]  [2400/6250]  eta: 0:17:15  lr: 0.000017  grad: 0.1872 (0.3010)  loss: 0.7835 (0.7840)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [77]  [2500/6250]  eta: 0:16:48  lr: 0.000017  grad: 0.1922 (0.3001)  loss: 0.7864 (0.7839)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [77]  [2600/6250]  eta: 0:16:21  lr: 0.000017  grad: 0.1854 (0.3025)  loss: 0.7827 (0.7839)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [77]  [2700/6250]  eta: 0:15:54  lr: 0.000017  grad: 0.1938 (0.3019)  loss: 0.7792 (0.7838)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [77]  [2800/6250]  eta: 0:15:27  lr: 0.000017  grad: 0.1770 (0.3007)  loss: 0.7878 (0.7838)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [77]  [2900/6250]  eta: 0:15:01  lr: 0.000017  grad: 0.1906 (0.3028)  loss: 0.7825 (0.7838)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [77]  [3000/6250]  eta: 0:14:33  lr: 0.000017  grad: 0.1867 (0.3050)  loss: 0.7811 (0.7837)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [77]  [3100/6250]  eta: 0:14:06  lr: 0.000017  grad: 0.1911 (0.3058)  loss: 0.7788 (0.7837)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [77]  [3200/6250]  eta: 0:13:39  lr: 0.000017  grad: 0.1796 (0.3087)  loss: 0.7840 (0.7836)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [77]  [3300/6250]  eta: 0:13:12  lr: 0.000016  grad: 0.1974 (0.3124)  loss: 0.7794 (0.7836)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [77]  [3400/6250]  eta: 0:12:45  lr: 0.000016  grad: 0.1881 (0.3138)  loss: 0.7782 (0.7835)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [77]  [3500/6250]  eta: 0:12:18  lr: 0.000016  grad: 0.1938 (0.3153)  loss: 0.7791 (0.7835)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [77]  [3600/6250]  eta: 0:11:51  lr: 0.000016  grad: 0.1977 (0.3182)  loss: 0.7874 (0.7835)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [77]  [3700/6250]  eta: 0:11:24  lr: 0.000016  grad: 0.2172 (0.3206)  loss: 0.7779 (0.7834)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [77]  [3800/6250]  eta: 0:10:57  lr: 0.000016  grad: 0.1841 (0.3218)  loss: 0.7806 (0.7833)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [77]  [3900/6250]  eta: 0:10:30  lr: 0.000016  grad: 0.1796 (0.3213)  loss: 0.7818 (0.7832)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [77]  [4000/6250]  eta: 0:10:04  lr: 0.000016  grad: 0.2103 (0.3212)  loss: 0.7770 (0.7831)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [4100/6250]  eta: 0:09:37  lr: 0.000016  grad: 0.1950 (0.3214)  loss: 0.7802 (0.7829)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [77]  [4200/6250]  eta: 0:09:10  lr: 0.000016  grad: 0.1941 (0.3224)  loss: 0.7775 (0.7828)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [77]  [4300/6250]  eta: 0:08:43  lr: 0.000016  grad: 0.2096 (0.3251)  loss: 0.7779 (0.7826)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [77]  [4400/6250]  eta: 0:08:16  lr: 0.000016  grad: 0.1982 (0.3252)  loss: 0.7769 (0.7825)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [77]  [4500/6250]  eta: 0:07:49  lr: 0.000016  grad: 0.2452 (0.3260)  loss: 0.7755 (0.7823)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [77]  [4600/6250]  eta: 0:07:22  lr: 0.000016  grad: 0.2279 (0.3272)  loss: 0.7724 (0.7821)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [4700/6250]  eta: 0:06:55  lr: 0.000016  grad: 0.1985 (0.3356)  loss: 0.7835 (0.7820)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [77]  [4800/6250]  eta: 0:06:28  lr: 0.000016  grad: 0.1802 (0.3348)  loss: 0.7824 (0.7819)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [77]  [4900/6250]  eta: 0:06:02  lr: 0.000016  grad: 0.2025 (0.3347)  loss: 0.7817 (0.7818)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [5000/6250]  eta: 0:05:35  lr: 0.000016  grad: 0.1788 (0.3359)  loss: 0.7793 (0.7817)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [77]  [5100/6250]  eta: 0:05:08  lr: 0.000016  grad: 0.2019 (0.3363)  loss: 0.7683 (0.7816)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [77]  [5200/6250]  eta: 0:04:41  lr: 0.000016  grad: 0.1804 (0.3373)  loss: 0.7906 (0.7816)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [77]  [5300/6250]  eta: 0:04:14  lr: 0.000016  grad: 0.1891 (0.3384)  loss: 0.7804 (0.7816)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [77]  [5400/6250]  eta: 0:03:47  lr: 0.000016  grad: 0.2030 (0.3383)  loss: 0.7765 (0.7816)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [77]  [5500/6250]  eta: 0:03:21  lr: 0.000016  grad: 0.1948 (0.3372)  loss: 0.7820 (0.7815)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [5600/6250]  eta: 0:02:54  lr: 0.000016  grad: 0.1758 (0.3375)  loss: 0.7808 (0.7815)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [77]  [5700/6250]  eta: 0:02:27  lr: 0.000016  grad: 0.1837 (0.3366)  loss: 0.7760 (0.7814)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [5800/6250]  eta: 0:02:00  lr: 0.000016  grad: 0.1924 (0.3373)  loss: 0.7780 (0.7813)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [77]  [5900/6250]  eta: 0:01:33  lr: 0.000016  grad: 0.1999 (0.3397)  loss: 0.7824 (0.7813)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [77]  [6000/6250]  eta: 0:01:06  lr: 0.000016  grad: 0.1993 (0.3416)  loss: 0.7827 (0.7812)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [77]  [6100/6250]  eta: 0:00:40  lr: 0.000016  grad: 0.2190 (0.3413)  loss: 0.7757 (0.7811)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [77]  [6200/6250]  eta: 0:00:13  lr: 0.000016  grad: 0.2049 (0.3404)  loss: 0.7707 (0.7811)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [77]  [6249/6250]  eta: 0:00:00  lr: 0.000016  grad: 0.2006 (0.3406)  loss: 0.7651 (0.7810)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [77] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000016  grad: 0.2006 (0.3406)  loss: 0.7651 (0.7810)
Eval (hcp-train-subset): [77]  [ 0/62]  eta: 0:05:07  loss: 0.8078 (0.8078)  time: 4.9547  data: 4.8717  max mem: 26157
Eval (hcp-train-subset): [77]  [61/62]  eta: 0:00:00  loss: 0.7969 (0.8010)  time: 0.1178  data: 0.0354  max mem: 26157
Eval (hcp-train-subset): [77] Total time: 0:00:11 (0.1847 s / it)
Averaged stats (hcp-train-subset): loss: 0.7969 (0.8010)
Making plots (hcp-train-subset): example=12
Eval (hcp-val): [77]  [ 0/62]  eta: 0:04:13  loss: 0.8202 (0.8202)  time: 4.0936  data: 4.0105  max mem: 26157
Eval (hcp-val): [77]  [61/62]  eta: 0:00:00  loss: 0.8227 (0.8242)  time: 0.0924  data: 0.0082  max mem: 26157
Eval (hcp-val): [77] Total time: 0:00:11 (0.1782 s / it)
Averaged stats (hcp-val): loss: 0.8227 (0.8242)
Making plots (hcp-val): example=43
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [78]  [   0/6250]  eta: 8:44:04  lr: 0.000016  grad: 0.2683 (0.2683)  loss: 0.8228 (0.8228)  time: 5.0311  data: 4.7617  max mem: 26157
Train: [78]  [ 100/6250]  eta: 0:32:17  lr: 0.000016  grad: 0.2510 (0.5075)  loss: 0.7911 (0.7919)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [78]  [ 200/6250]  eta: 0:29:21  lr: 0.000016  grad: 0.2578 (0.4132)  loss: 0.7753 (0.7822)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [78]  [ 300/6250]  eta: 0:28:04  lr: 0.000016  grad: 0.2032 (0.3727)  loss: 0.7885 (0.7807)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [78]  [ 400/6250]  eta: 0:27:12  lr: 0.000016  grad: 0.2109 (0.3697)  loss: 0.7758 (0.7800)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [78]  [ 500/6250]  eta: 0:26:31  lr: 0.000016  grad: 0.2083 (0.3750)  loss: 0.7954 (0.7810)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [78]  [ 600/6250]  eta: 0:25:54  lr: 0.000016  grad: 0.1913 (0.3577)  loss: 0.7927 (0.7825)  time: 0.2660  data: 0.0002  max mem: 26157
Train: [78]  [ 700/6250]  eta: 0:25:20  lr: 0.000016  grad: 0.1888 (0.3428)  loss: 0.7886 (0.7839)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [78]  [ 800/6250]  eta: 0:24:48  lr: 0.000016  grad: 0.1866 (0.3486)  loss: 0.7926 (0.7850)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [78]  [ 900/6250]  eta: 0:24:17  lr: 0.000016  grad: 0.1849 (0.3360)  loss: 0.7888 (0.7855)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [78]  [1000/6250]  eta: 0:23:47  lr: 0.000016  grad: 0.1786 (0.3315)  loss: 0.7879 (0.7860)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [78]  [1100/6250]  eta: 0:23:17  lr: 0.000016  grad: 0.1896 (0.3328)  loss: 0.7857 (0.7861)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [78]  [1200/6250]  eta: 0:22:49  lr: 0.000016  grad: 0.1790 (0.3298)  loss: 0.7771 (0.7862)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [78]  [1300/6250]  eta: 0:22:20  lr: 0.000016  grad: 0.1942 (0.3319)  loss: 0.7790 (0.7861)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [78]  [1400/6250]  eta: 0:21:52  lr: 0.000016  grad: 0.1848 (0.3371)  loss: 0.7786 (0.7858)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [78]  [1500/6250]  eta: 0:21:24  lr: 0.000015  grad: 0.1800 (0.3327)  loss: 0.7800 (0.7855)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [78]  [1600/6250]  eta: 0:20:56  lr: 0.000015  grad: 0.1942 (0.3280)  loss: 0.7789 (0.7852)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [78]  [1700/6250]  eta: 0:20:28  lr: 0.000015  grad: 0.1916 (0.3278)  loss: 0.7833 (0.7849)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [78]  [1800/6250]  eta: 0:20:00  lr: 0.000015  grad: 0.1747 (0.3230)  loss: 0.7869 (0.7849)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [78]  [1900/6250]  eta: 0:19:32  lr: 0.000015  grad: 0.1993 (0.3214)  loss: 0.7853 (0.7848)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [78]  [2000/6250]  eta: 0:19:05  lr: 0.000015  grad: 0.1876 (0.3179)  loss: 0.7816 (0.7849)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [78]  [2100/6250]  eta: 0:18:37  lr: 0.000015  grad: 0.1905 (0.3169)  loss: 0.7822 (0.7847)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [78]  [2200/6250]  eta: 0:18:10  lr: 0.000015  grad: 0.1803 (0.3149)  loss: 0.7917 (0.7848)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [78]  [2300/6250]  eta: 0:17:43  lr: 0.000015  grad: 0.1893 (0.3180)  loss: 0.7897 (0.7847)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [78]  [2400/6250]  eta: 0:17:15  lr: 0.000015  grad: 0.1812 (0.3187)  loss: 0.7881 (0.7847)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [78]  [2500/6250]  eta: 0:16:48  lr: 0.000015  grad: 0.1959 (0.3175)  loss: 0.7811 (0.7846)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [78]  [2600/6250]  eta: 0:16:21  lr: 0.000015  grad: 0.1914 (0.3168)  loss: 0.7774 (0.7845)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [78]  [2700/6250]  eta: 0:15:54  lr: 0.000015  grad: 0.1883 (0.3151)  loss: 0.7848 (0.7844)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [78]  [2800/6250]  eta: 0:15:27  lr: 0.000015  grad: 0.1863 (0.3174)  loss: 0.7919 (0.7845)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [78]  [2900/6250]  eta: 0:15:00  lr: 0.000015  grad: 0.1879 (0.3172)  loss: 0.7785 (0.7844)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [78]  [3000/6250]  eta: 0:14:33  lr: 0.000015  grad: 0.1863 (0.3146)  loss: 0.7854 (0.7843)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [78]  [3100/6250]  eta: 0:14:05  lr: 0.000015  grad: 0.1909 (0.3179)  loss: 0.7852 (0.7842)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [78]  [3200/6250]  eta: 0:13:38  lr: 0.000015  grad: 0.1868 (0.3180)  loss: 0.7883 (0.7841)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [78]  [3300/6250]  eta: 0:13:11  lr: 0.000015  grad: 0.1827 (0.3211)  loss: 0.7827 (0.7840)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [78]  [3400/6250]  eta: 0:12:44  lr: 0.000015  grad: 0.1962 (0.3203)  loss: 0.7794 (0.7839)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [78]  [3500/6250]  eta: 0:12:18  lr: 0.000015  grad: 0.2015 (0.3209)  loss: 0.7902 (0.7837)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [78]  [3600/6250]  eta: 0:11:51  lr: 0.000015  grad: 0.1991 (0.3218)  loss: 0.7814 (0.7836)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [78]  [3700/6250]  eta: 0:11:24  lr: 0.000015  grad: 0.2202 (0.3191)  loss: 0.7808 (0.7835)  time: 0.2658  data: 0.0001  max mem: 26157
Train: [78]  [3800/6250]  eta: 0:10:57  lr: 0.000015  grad: 0.1808 (0.3184)  loss: 0.7785 (0.7834)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [78]  [3900/6250]  eta: 0:10:30  lr: 0.000015  grad: 0.1985 (0.3180)  loss: 0.7857 (0.7833)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [78]  [4000/6250]  eta: 0:10:03  lr: 0.000015  grad: 0.1980 (0.3166)  loss: 0.7843 (0.7831)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [78]  [4100/6250]  eta: 0:09:36  lr: 0.000015  grad: 0.1916 (0.3164)  loss: 0.7848 (0.7831)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [78]  [4200/6250]  eta: 0:09:09  lr: 0.000015  grad: 0.1839 (0.3153)  loss: 0.7802 (0.7829)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [78]  [4300/6250]  eta: 0:08:42  lr: 0.000015  grad: 0.1990 (0.3215)  loss: 0.7747 (0.7829)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [78]  [4400/6250]  eta: 0:08:15  lr: 0.000015  grad: 0.1899 (0.3199)  loss: 0.7788 (0.7828)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [78]  [4500/6250]  eta: 0:07:49  lr: 0.000015  grad: 0.1861 (0.3187)  loss: 0.7774 (0.7827)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [78]  [4600/6250]  eta: 0:07:22  lr: 0.000015  grad: 0.2011 (0.3180)  loss: 0.7694 (0.7826)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [78]  [4700/6250]  eta: 0:06:55  lr: 0.000015  grad: 0.1873 (0.3184)  loss: 0.7795 (0.7825)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [78]  [4800/6250]  eta: 0:06:28  lr: 0.000015  grad: 0.2014 (0.3180)  loss: 0.7778 (0.7824)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [78]  [4900/6250]  eta: 0:06:01  lr: 0.000015  grad: 0.1937 (0.3201)  loss: 0.7890 (0.7823)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [78]  [5000/6250]  eta: 0:05:34  lr: 0.000015  grad: 0.1883 (0.3203)  loss: 0.7804 (0.7822)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [78]  [5100/6250]  eta: 0:05:08  lr: 0.000015  grad: 0.1861 (0.3206)  loss: 0.7832 (0.7822)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [78]  [5200/6250]  eta: 0:04:41  lr: 0.000015  grad: 0.1874 (0.3198)  loss: 0.7813 (0.7822)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [78]  [5300/6250]  eta: 0:04:14  lr: 0.000015  grad: 0.1961 (0.3223)  loss: 0.7852 (0.7821)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [78]  [5400/6250]  eta: 0:03:47  lr: 0.000015  grad: 0.1963 (0.3243)  loss: 0.7772 (0.7820)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [78]  [5500/6250]  eta: 0:03:20  lr: 0.000015  grad: 0.1948 (0.3253)  loss: 0.7799 (0.7820)  time: 0.2657  data: 0.0001  max mem: 26157
Train: [78]  [5600/6250]  eta: 0:02:54  lr: 0.000015  grad: 0.1911 (0.3241)  loss: 0.7831 (0.7819)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [78]  [5700/6250]  eta: 0:02:27  lr: 0.000015  grad: 0.2098 (0.3244)  loss: 0.7790 (0.7818)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [78]  [5800/6250]  eta: 0:02:00  lr: 0.000015  grad: 0.1945 (0.3243)  loss: 0.7712 (0.7817)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [78]  [5900/6250]  eta: 0:01:33  lr: 0.000015  grad: 0.1917 (0.3242)  loss: 0.7782 (0.7816)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [78]  [6000/6250]  eta: 0:01:06  lr: 0.000015  grad: 0.1932 (0.3268)  loss: 0.7744 (0.7815)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [78]  [6100/6250]  eta: 0:00:40  lr: 0.000015  grad: 0.1905 (0.3269)  loss: 0.7791 (0.7814)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [78]  [6200/6250]  eta: 0:00:13  lr: 0.000014  grad: 0.1871 (0.3279)  loss: 0.7830 (0.7813)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [78]  [6249/6250]  eta: 0:00:00  lr: 0.000014  grad: 0.1939 (0.3280)  loss: 0.7732 (0.7812)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [78] Total time: 0:27:56 (0.2682 s / it)
Averaged stats: lr: 0.000014  grad: 0.1939 (0.3280)  loss: 0.7732 (0.7812)
Eval (hcp-train-subset): [78]  [ 0/62]  eta: 0:04:37  loss: 0.8102 (0.8102)  time: 4.4824  data: 4.3991  max mem: 26157
Eval (hcp-train-subset): [78]  [61/62]  eta: 0:00:00  loss: 0.7954 (0.7999)  time: 0.0906  data: 0.0065  max mem: 26157
Eval (hcp-train-subset): [78] Total time: 0:00:10 (0.1672 s / it)
Averaged stats (hcp-train-subset): loss: 0.7954 (0.7999)
Making plots (hcp-train-subset): example=62
Eval (hcp-val): [78]  [ 0/62]  eta: 0:04:12  loss: 0.8197 (0.8197)  time: 4.0784  data: 3.9952  max mem: 26157
Eval (hcp-val): [78]  [61/62]  eta: 0:00:00  loss: 0.8227 (0.8245)  time: 0.0893  data: 0.0072  max mem: 26157
Eval (hcp-val): [78] Total time: 0:00:10 (0.1674 s / it)
Averaged stats (hcp-val): loss: 0.8227 (0.8245)
Making plots (hcp-val): example=29
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [79]  [   0/6250]  eta: 6:14:29  lr: 0.000014  grad: 0.1601 (0.1601)  loss: 0.8393 (0.8393)  time: 3.5951  data: 3.2735  max mem: 26157
Train: [79]  [ 100/6250]  eta: 0:32:02  lr: 0.000014  grad: 0.2150 (0.3866)  loss: 0.7990 (0.7917)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [79]  [ 200/6250]  eta: 0:29:17  lr: 0.000014  grad: 0.2119 (0.3676)  loss: 0.7931 (0.7890)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [79]  [ 300/6250]  eta: 0:28:03  lr: 0.000014  grad: 0.1789 (0.3233)  loss: 0.7914 (0.7872)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [79]  [ 400/6250]  eta: 0:27:13  lr: 0.000014  grad: 0.1881 (0.3375)  loss: 0.7780 (0.7869)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [79]  [ 500/6250]  eta: 0:26:32  lr: 0.000014  grad: 0.2059 (0.3580)  loss: 0.7774 (0.7863)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [79]  [ 600/6250]  eta: 0:25:55  lr: 0.000014  grad: 0.1894 (0.3460)  loss: 0.7799 (0.7854)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [79]  [ 700/6250]  eta: 0:25:21  lr: 0.000014  grad: 0.1890 (0.3329)  loss: 0.7828 (0.7847)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [79]  [ 800/6250]  eta: 0:24:49  lr: 0.000014  grad: 0.2060 (0.3381)  loss: 0.7721 (0.7837)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [79]  [ 900/6250]  eta: 0:24:18  lr: 0.000014  grad: 0.2025 (0.3343)  loss: 0.7742 (0.7828)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [79]  [1000/6250]  eta: 0:23:48  lr: 0.000014  grad: 0.1932 (0.3296)  loss: 0.7792 (0.7820)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [79]  [1100/6250]  eta: 0:23:18  lr: 0.000014  grad: 0.1873 (0.3239)  loss: 0.7732 (0.7815)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [79]  [1200/6250]  eta: 0:22:49  lr: 0.000014  grad: 0.1972 (0.3243)  loss: 0.7765 (0.7810)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [79]  [1300/6250]  eta: 0:22:20  lr: 0.000014  grad: 0.1891 (0.3235)  loss: 0.7782 (0.7805)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [79]  [1400/6250]  eta: 0:21:52  lr: 0.000014  grad: 0.1918 (0.3228)  loss: 0.7736 (0.7801)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [79]  [1500/6250]  eta: 0:21:24  lr: 0.000014  grad: 0.1951 (0.3222)  loss: 0.7786 (0.7800)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [79]  [1600/6250]  eta: 0:20:56  lr: 0.000014  grad: 0.1874 (0.3230)  loss: 0.7734 (0.7799)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [79]  [1700/6250]  eta: 0:20:28  lr: 0.000014  grad: 0.2157 (0.3295)  loss: 0.7812 (0.7799)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [79]  [1800/6250]  eta: 0:20:00  lr: 0.000014  grad: 0.2053 (0.3271)  loss: 0.7793 (0.7798)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [79]  [1900/6250]  eta: 0:19:32  lr: 0.000014  grad: 0.1794 (0.3255)  loss: 0.7805 (0.7798)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [79]  [2000/6250]  eta: 0:19:05  lr: 0.000014  grad: 0.1887 (0.3317)  loss: 0.7811 (0.7796)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [79]  [2100/6250]  eta: 0:18:37  lr: 0.000014  grad: 0.2080 (0.3295)  loss: 0.7803 (0.7796)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [79]  [2200/6250]  eta: 0:18:10  lr: 0.000014  grad: 0.1863 (0.3299)  loss: 0.7876 (0.7798)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [79]  [2300/6250]  eta: 0:17:43  lr: 0.000014  grad: 0.1790 (0.3307)  loss: 0.7805 (0.7799)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [79]  [2400/6250]  eta: 0:17:16  lr: 0.000014  grad: 0.1866 (0.3314)  loss: 0.7791 (0.7799)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [79]  [2500/6250]  eta: 0:16:48  lr: 0.000014  grad: 0.1968 (0.3328)  loss: 0.7720 (0.7799)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [79]  [2600/6250]  eta: 0:16:21  lr: 0.000014  grad: 0.1965 (0.3312)  loss: 0.7709 (0.7799)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [79]  [2700/6250]  eta: 0:15:54  lr: 0.000014  grad: 0.1811 (0.3335)  loss: 0.7798 (0.7799)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [79]  [2800/6250]  eta: 0:15:27  lr: 0.000014  grad: 0.1990 (0.3373)  loss: 0.7826 (0.7799)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [79]  [2900/6250]  eta: 0:15:00  lr: 0.000014  grad: 0.1833 (0.3385)  loss: 0.7857 (0.7799)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [79]  [3000/6250]  eta: 0:14:33  lr: 0.000014  grad: 0.2078 (0.3397)  loss: 0.7833 (0.7798)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [79]  [3100/6250]  eta: 0:14:06  lr: 0.000014  grad: 0.2112 (0.3396)  loss: 0.7857 (0.7799)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [79]  [3200/6250]  eta: 0:13:39  lr: 0.000014  grad: 0.1809 (0.3375)  loss: 0.7766 (0.7800)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [79]  [3300/6250]  eta: 0:13:12  lr: 0.000014  grad: 0.1940 (0.3364)  loss: 0.7770 (0.7800)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [79]  [3400/6250]  eta: 0:12:45  lr: 0.000014  grad: 0.2049 (0.3367)  loss: 0.7800 (0.7800)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [79]  [3500/6250]  eta: 0:12:18  lr: 0.000014  grad: 0.1859 (0.3342)  loss: 0.7821 (0.7799)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [79]  [3600/6250]  eta: 0:11:51  lr: 0.000014  grad: 0.1917 (0.3331)  loss: 0.7790 (0.7799)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [79]  [3700/6250]  eta: 0:11:24  lr: 0.000014  grad: 0.1874 (0.3346)  loss: 0.7873 (0.7799)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [79]  [3800/6250]  eta: 0:10:57  lr: 0.000014  grad: 0.1917 (0.3317)  loss: 0.7704 (0.7799)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [79]  [3900/6250]  eta: 0:10:30  lr: 0.000014  grad: 0.1886 (0.3306)  loss: 0.7833 (0.7799)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [79]  [4000/6250]  eta: 0:10:03  lr: 0.000014  grad: 0.1914 (0.3296)  loss: 0.7822 (0.7798)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [79]  [4100/6250]  eta: 0:09:36  lr: 0.000014  grad: 0.2058 (0.3334)  loss: 0.7729 (0.7797)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [79]  [4200/6250]  eta: 0:09:09  lr: 0.000014  grad: 0.2019 (0.3326)  loss: 0.7791 (0.7797)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [79]  [4300/6250]  eta: 0:08:43  lr: 0.000014  grad: 0.1895 (0.3333)  loss: 0.7862 (0.7796)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [79]  [4400/6250]  eta: 0:08:16  lr: 0.000014  grad: 0.1838 (0.3311)  loss: 0.7812 (0.7795)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [79]  [4500/6250]  eta: 0:07:49  lr: 0.000014  grad: 0.1970 (0.3305)  loss: 0.7741 (0.7796)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [79]  [4600/6250]  eta: 0:07:22  lr: 0.000014  grad: 0.1823 (0.3306)  loss: 0.7802 (0.7795)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [79]  [4700/6250]  eta: 0:06:55  lr: 0.000013  grad: 0.1860 (0.3310)  loss: 0.7779 (0.7795)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [79]  [4800/6250]  eta: 0:06:28  lr: 0.000013  grad: 0.1922 (0.3305)  loss: 0.7791 (0.7795)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [79]  [4900/6250]  eta: 0:06:01  lr: 0.000013  grad: 0.1969 (0.3314)  loss: 0.7795 (0.7795)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [79]  [5000/6250]  eta: 0:05:35  lr: 0.000013  grad: 0.1829 (0.3297)  loss: 0.7822 (0.7795)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [79]  [5100/6250]  eta: 0:05:08  lr: 0.000013  grad: 0.2023 (0.3281)  loss: 0.7778 (0.7795)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [79]  [5200/6250]  eta: 0:04:41  lr: 0.000013  grad: 0.1941 (0.3297)  loss: 0.7794 (0.7796)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [79]  [5300/6250]  eta: 0:04:14  lr: 0.000013  grad: 0.1793 (0.3292)  loss: 0.7878 (0.7796)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [79]  [5400/6250]  eta: 0:03:47  lr: 0.000013  grad: 0.2134 (0.3285)  loss: 0.7826 (0.7796)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [79]  [5500/6250]  eta: 0:03:20  lr: 0.000013  grad: 0.1843 (0.3267)  loss: 0.7853 (0.7797)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [79]  [5600/6250]  eta: 0:02:54  lr: 0.000013  grad: 0.1824 (0.3267)  loss: 0.7838 (0.7798)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [79]  [5700/6250]  eta: 0:02:27  lr: 0.000013  grad: 0.1970 (0.3262)  loss: 0.7805 (0.7798)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [79]  [5800/6250]  eta: 0:02:00  lr: 0.000013  grad: 0.1833 (0.3258)  loss: 0.7784 (0.7798)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [79]  [5900/6250]  eta: 0:01:33  lr: 0.000013  grad: 0.2205 (0.3254)  loss: 0.7742 (0.7797)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [79]  [6000/6250]  eta: 0:01:06  lr: 0.000013  grad: 0.2000 (0.3240)  loss: 0.7759 (0.7797)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [79]  [6100/6250]  eta: 0:00:40  lr: 0.000013  grad: 0.2056 (0.3241)  loss: 0.7794 (0.7797)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [79]  [6200/6250]  eta: 0:00:13  lr: 0.000013  grad: 0.1906 (0.3237)  loss: 0.7747 (0.7796)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [79]  [6249/6250]  eta: 0:00:00  lr: 0.000013  grad: 0.1884 (0.3232)  loss: 0.7880 (0.7796)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [79] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000013  grad: 0.1884 (0.3232)  loss: 0.7880 (0.7796)
Eval (hcp-train-subset): [79]  [ 0/62]  eta: 0:04:47  loss: 0.8078 (0.8078)  time: 4.6416  data: 4.5583  max mem: 26157
Eval (hcp-train-subset): [79]  [61/62]  eta: 0:00:00  loss: 0.7960 (0.8004)  time: 0.0905  data: 0.0080  max mem: 26157
Eval (hcp-train-subset): [79] Total time: 0:00:10 (0.1722 s / it)
Averaged stats (hcp-train-subset): loss: 0.7960 (0.8004)
Making plots (hcp-train-subset): example=9
Eval (hcp-val): [79]  [ 0/62]  eta: 0:03:19  loss: 0.8186 (0.8186)  time: 3.2183  data: 3.1116  max mem: 26157
Eval (hcp-val): [79]  [61/62]  eta: 0:00:00  loss: 0.8226 (0.8238)  time: 0.0979  data: 0.0153  max mem: 26157
Eval (hcp-val): [79] Total time: 0:00:11 (0.1850 s / it)
Averaged stats (hcp-val): loss: 0.8226 (0.8238)
Making plots (hcp-val): example=37
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [80]  [   0/6250]  eta: 6:43:18  lr: 0.000013  grad: 0.1634 (0.1634)  loss: 0.8307 (0.8307)  time: 3.8717  data: 3.5267  max mem: 26157
Train: [80]  [ 100/6250]  eta: 0:32:03  lr: 0.000013  grad: 0.2215 (0.4047)  loss: 0.7985 (0.8101)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [80]  [ 200/6250]  eta: 0:29:16  lr: 0.000013  grad: 0.2431 (0.3340)  loss: 0.7884 (0.8015)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [80]  [ 300/6250]  eta: 0:28:00  lr: 0.000013  grad: 0.2161 (0.3169)  loss: 0.7918 (0.7975)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [80]  [ 400/6250]  eta: 0:27:10  lr: 0.000013  grad: 0.1776 (0.3194)  loss: 0.7961 (0.7961)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [80]  [ 500/6250]  eta: 0:26:29  lr: 0.000013  grad: 0.2100 (0.3055)  loss: 0.7800 (0.7950)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [80]  [ 600/6250]  eta: 0:25:53  lr: 0.000013  grad: 0.2024 (0.2982)  loss: 0.7842 (0.7940)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [80]  [ 700/6250]  eta: 0:25:19  lr: 0.000013  grad: 0.2118 (0.3085)  loss: 0.7859 (0.7932)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [80]  [ 800/6250]  eta: 0:24:47  lr: 0.000013  grad: 0.1895 (0.3037)  loss: 0.7827 (0.7921)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [80]  [ 900/6250]  eta: 0:24:16  lr: 0.000013  grad: 0.2037 (0.3210)  loss: 0.7748 (0.7911)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [80]  [1000/6250]  eta: 0:23:46  lr: 0.000013  grad: 0.1974 (0.3176)  loss: 0.7823 (0.7903)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [80]  [1100/6250]  eta: 0:23:17  lr: 0.000013  grad: 0.2046 (0.3165)  loss: 0.7720 (0.7894)  time: 0.2663  data: 0.0002  max mem: 26157
Train: [80]  [1200/6250]  eta: 0:22:48  lr: 0.000013  grad: 0.2016 (0.3177)  loss: 0.7754 (0.7887)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [80]  [1300/6250]  eta: 0:22:19  lr: 0.000013  grad: 0.1922 (0.3161)  loss: 0.7823 (0.7880)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [80]  [1400/6250]  eta: 0:21:51  lr: 0.000013  grad: 0.1878 (0.3181)  loss: 0.7839 (0.7876)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [80]  [1500/6250]  eta: 0:21:23  lr: 0.000013  grad: 0.2053 (0.3158)  loss: 0.7757 (0.7872)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [80]  [1600/6250]  eta: 0:20:55  lr: 0.000013  grad: 0.1964 (0.3112)  loss: 0.7769 (0.7867)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [80]  [1700/6250]  eta: 0:20:27  lr: 0.000013  grad: 0.2191 (0.3125)  loss: 0.7753 (0.7863)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [80]  [1800/6250]  eta: 0:19:59  lr: 0.000013  grad: 0.1895 (0.3121)  loss: 0.7768 (0.7859)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [80]  [1900/6250]  eta: 0:19:32  lr: 0.000013  grad: 0.1990 (0.3116)  loss: 0.7726 (0.7854)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [80]  [2000/6250]  eta: 0:19:04  lr: 0.000013  grad: 0.1867 (0.3083)  loss: 0.7785 (0.7850)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [80]  [2100/6250]  eta: 0:18:37  lr: 0.000013  grad: 0.1878 (0.3068)  loss: 0.7683 (0.7846)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [80]  [2200/6250]  eta: 0:18:09  lr: 0.000013  grad: 0.1962 (0.3089)  loss: 0.7855 (0.7844)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [80]  [2300/6250]  eta: 0:17:42  lr: 0.000013  grad: 0.1815 (0.3099)  loss: 0.7804 (0.7841)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [80]  [2400/6250]  eta: 0:17:15  lr: 0.000013  grad: 0.1792 (0.3081)  loss: 0.7835 (0.7839)  time: 0.2658  data: 0.0001  max mem: 26157
Train: [80]  [2500/6250]  eta: 0:16:48  lr: 0.000013  grad: 0.1905 (0.3066)  loss: 0.7850 (0.7838)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [80]  [2600/6250]  eta: 0:16:21  lr: 0.000013  grad: 0.1888 (0.3074)  loss: 0.7795 (0.7836)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [80]  [2700/6250]  eta: 0:15:54  lr: 0.000013  grad: 0.2156 (0.3128)  loss: 0.7831 (0.7834)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [80]  [2800/6250]  eta: 0:15:27  lr: 0.000013  grad: 0.1958 (0.3102)  loss: 0.7824 (0.7833)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [80]  [2900/6250]  eta: 0:15:00  lr: 0.000013  grad: 0.2138 (0.3094)  loss: 0.7766 (0.7831)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [80]  [3000/6250]  eta: 0:14:33  lr: 0.000013  grad: 0.1985 (0.3127)  loss: 0.7767 (0.7829)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [80]  [3100/6250]  eta: 0:14:06  lr: 0.000013  grad: 0.1939 (0.3115)  loss: 0.7790 (0.7827)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [80]  [3200/6250]  eta: 0:13:39  lr: 0.000013  grad: 0.1984 (0.3141)  loss: 0.7798 (0.7825)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [80]  [3300/6250]  eta: 0:13:12  lr: 0.000013  grad: 0.1888 (0.3180)  loss: 0.7757 (0.7823)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [80]  [3400/6250]  eta: 0:12:45  lr: 0.000012  grad: 0.1963 (0.3177)  loss: 0.7824 (0.7822)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [80]  [3500/6250]  eta: 0:12:18  lr: 0.000012  grad: 0.1890 (0.3185)  loss: 0.7698 (0.7820)  time: 0.2690  data: 0.0002  max mem: 26157
Train: [80]  [3600/6250]  eta: 0:11:51  lr: 0.000012  grad: 0.1956 (0.3179)  loss: 0.7796 (0.7819)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [80]  [3700/6250]  eta: 0:11:24  lr: 0.000012  grad: 0.1938 (0.3186)  loss: 0.7762 (0.7819)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [80]  [3800/6250]  eta: 0:10:57  lr: 0.000012  grad: 0.2032 (0.3254)  loss: 0.7804 (0.7818)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [80]  [3900/6250]  eta: 0:10:30  lr: 0.000012  grad: 0.1896 (0.3242)  loss: 0.7746 (0.7818)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [80]  [4000/6250]  eta: 0:10:03  lr: 0.000012  grad: 0.2124 (0.3257)  loss: 0.7850 (0.7818)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [80]  [4100/6250]  eta: 0:09:37  lr: 0.000012  grad: 0.2129 (0.3265)  loss: 0.7789 (0.7818)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [80]  [4200/6250]  eta: 0:09:10  lr: 0.000012  grad: 0.2143 (0.3294)  loss: 0.7840 (0.7818)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [80]  [4300/6250]  eta: 0:08:43  lr: 0.000012  grad: 0.1969 (0.3315)  loss: 0.7841 (0.7818)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [80]  [4400/6250]  eta: 0:08:16  lr: 0.000012  grad: 0.1961 (0.3315)  loss: 0.7933 (0.7819)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [80]  [4500/6250]  eta: 0:07:49  lr: 0.000012  grad: 0.1981 (0.3344)  loss: 0.7868 (0.7819)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [80]  [4600/6250]  eta: 0:07:22  lr: 0.000012  grad: 0.1980 (0.3334)  loss: 0.7776 (0.7820)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [80]  [4700/6250]  eta: 0:06:55  lr: 0.000012  grad: 0.2058 (0.3319)  loss: 0.7843 (0.7819)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [80]  [4800/6250]  eta: 0:06:28  lr: 0.000012  grad: 0.1989 (0.3312)  loss: 0.7864 (0.7819)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [80]  [4900/6250]  eta: 0:06:02  lr: 0.000012  grad: 0.1879 (0.3306)  loss: 0.7858 (0.7820)  time: 0.2663  data: 0.0002  max mem: 26157
Train: [80]  [5000/6250]  eta: 0:05:35  lr: 0.000012  grad: 0.1999 (0.3290)  loss: 0.7782 (0.7820)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [80]  [5100/6250]  eta: 0:05:08  lr: 0.000012  grad: 0.1935 (0.3274)  loss: 0.7793 (0.7819)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [80]  [5200/6250]  eta: 0:04:41  lr: 0.000012  grad: 0.1954 (0.3263)  loss: 0.7865 (0.7819)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [80]  [5300/6250]  eta: 0:04:14  lr: 0.000012  grad: 0.1935 (0.3277)  loss: 0.7845 (0.7820)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [80]  [5400/6250]  eta: 0:03:47  lr: 0.000012  grad: 0.1907 (0.3269)  loss: 0.7799 (0.7819)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [80]  [5500/6250]  eta: 0:03:21  lr: 0.000012  grad: 0.2025 (0.3264)  loss: 0.7897 (0.7819)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [80]  [5600/6250]  eta: 0:02:54  lr: 0.000012  grad: 0.1907 (0.3264)  loss: 0.7791 (0.7820)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [80]  [5700/6250]  eta: 0:02:27  lr: 0.000012  grad: 0.1935 (0.3258)  loss: 0.7827 (0.7819)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [80]  [5800/6250]  eta: 0:02:00  lr: 0.000012  grad: 0.1963 (0.3258)  loss: 0.7862 (0.7820)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [80]  [5900/6250]  eta: 0:01:33  lr: 0.000012  grad: 0.1946 (0.3247)  loss: 0.7767 (0.7819)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [80]  [6000/6250]  eta: 0:01:07  lr: 0.000012  grad: 0.2042 (0.3267)  loss: 0.7814 (0.7819)  time: 0.2661  data: 0.0002  max mem: 26157
Train: [80]  [6100/6250]  eta: 0:00:40  lr: 0.000012  grad: 0.2001 (0.3261)  loss: 0.7769 (0.7819)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [80]  [6200/6250]  eta: 0:00:13  lr: 0.000012  grad: 0.2021 (0.3258)  loss: 0.7824 (0.7818)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [80]  [6249/6250]  eta: 0:00:00  lr: 0.000012  grad: 0.2020 (0.3259)  loss: 0.7747 (0.7818)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [80] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000012  grad: 0.2020 (0.3259)  loss: 0.7747 (0.7818)
Eval (hcp-train-subset): [80]  [ 0/62]  eta: 0:03:25  loss: 0.8126 (0.8126)  time: 3.3075  data: 3.2103  max mem: 26157
Eval (hcp-train-subset): [80]  [61/62]  eta: 0:00:00  loss: 0.8004 (0.8009)  time: 0.0824  data: 0.0001  max mem: 26157
Eval (hcp-train-subset): [80] Total time: 0:00:10 (0.1691 s / it)
Averaged stats (hcp-train-subset): loss: 0.8004 (0.8009)
Making plots (hcp-train-subset): example=5
Eval (hcp-val): [80]  [ 0/62]  eta: 0:02:34  loss: 0.8205 (0.8205)  time: 2.4885  data: 2.3935  max mem: 26157
Eval (hcp-val): [80]  [61/62]  eta: 0:00:00  loss: 0.8231 (0.8240)  time: 0.0838  data: 0.0001  max mem: 26157
Eval (hcp-val): [80] Total time: 0:00:10 (0.1660 s / it)
Averaged stats (hcp-val): loss: 0.8231 (0.8240)
Making plots (hcp-val): example=58
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [81]  [   0/6250]  eta: 6:38:27  lr: 0.000012  grad: 0.5728 (0.5728)  loss: 0.8052 (0.8052)  time: 3.8252  data: 3.5196  max mem: 26157
Train: [81]  [ 100/6250]  eta: 0:32:15  lr: 0.000012  grad: 0.2515 (0.3394)  loss: 0.7844 (0.7893)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [81]  [ 200/6250]  eta: 0:29:19  lr: 0.000012  grad: 0.2061 (0.3344)  loss: 0.7958 (0.7902)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [81]  [ 300/6250]  eta: 0:28:04  lr: 0.000012  grad: 0.1983 (0.2991)  loss: 0.7889 (0.7895)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [81]  [ 400/6250]  eta: 0:27:12  lr: 0.000012  grad: 0.1956 (0.2919)  loss: 0.7868 (0.7886)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [81]  [ 500/6250]  eta: 0:26:31  lr: 0.000012  grad: 0.1852 (0.2980)  loss: 0.7875 (0.7881)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [81]  [ 600/6250]  eta: 0:25:55  lr: 0.000012  grad: 0.2011 (0.3080)  loss: 0.7829 (0.7877)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [81]  [ 700/6250]  eta: 0:25:22  lr: 0.000012  grad: 0.1967 (0.3021)  loss: 0.7915 (0.7876)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [81]  [ 800/6250]  eta: 0:24:50  lr: 0.000012  grad: 0.1906 (0.3058)  loss: 0.7794 (0.7871)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [81]  [ 900/6250]  eta: 0:24:19  lr: 0.000012  grad: 0.1923 (0.3063)  loss: 0.7855 (0.7866)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [81]  [1000/6250]  eta: 0:23:49  lr: 0.000012  grad: 0.1971 (0.3023)  loss: 0.7802 (0.7858)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [81]  [1100/6250]  eta: 0:23:20  lr: 0.000012  grad: 0.2008 (0.3093)  loss: 0.7718 (0.7851)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [81]  [1200/6250]  eta: 0:22:50  lr: 0.000012  grad: 0.1908 (0.3112)  loss: 0.7776 (0.7842)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [81]  [1300/6250]  eta: 0:22:22  lr: 0.000012  grad: 0.1913 (0.3092)  loss: 0.7815 (0.7838)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [81]  [1400/6250]  eta: 0:21:53  lr: 0.000012  grad: 0.1995 (0.3144)  loss: 0.7769 (0.7832)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [81]  [1500/6250]  eta: 0:21:25  lr: 0.000012  grad: 0.1867 (0.3133)  loss: 0.7799 (0.7828)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [81]  [1600/6250]  eta: 0:20:57  lr: 0.000012  grad: 0.1940 (0.3118)  loss: 0.7765 (0.7825)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [81]  [1700/6250]  eta: 0:20:29  lr: 0.000012  grad: 0.2057 (0.3151)  loss: 0.7846 (0.7821)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [81]  [1800/6250]  eta: 0:20:01  lr: 0.000012  grad: 0.1871 (0.3234)  loss: 0.7696 (0.7819)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [81]  [1900/6250]  eta: 0:19:33  lr: 0.000012  grad: 0.2024 (0.3250)  loss: 0.7754 (0.7816)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [81]  [2000/6250]  eta: 0:19:06  lr: 0.000012  grad: 0.2150 (0.3232)  loss: 0.7735 (0.7812)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [81]  [2100/6250]  eta: 0:18:38  lr: 0.000012  grad: 0.2056 (0.3209)  loss: 0.7796 (0.7810)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [81]  [2200/6250]  eta: 0:18:11  lr: 0.000012  grad: 0.1969 (0.3192)  loss: 0.7714 (0.7808)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [81]  [2300/6250]  eta: 0:17:43  lr: 0.000011  grad: 0.1975 (0.3230)  loss: 0.7814 (0.7807)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [81]  [2400/6250]  eta: 0:17:16  lr: 0.000011  grad: 0.2124 (0.3276)  loss: 0.7780 (0.7806)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [81]  [2500/6250]  eta: 0:16:49  lr: 0.000011  grad: 0.1971 (0.3266)  loss: 0.7777 (0.7806)  time: 0.2663  data: 0.0002  max mem: 26157
Train: [81]  [2600/6250]  eta: 0:16:22  lr: 0.000011  grad: 0.1890 (0.3269)  loss: 0.7732 (0.7805)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [81]  [2700/6250]  eta: 0:15:54  lr: 0.000011  grad: 0.1852 (0.3273)  loss: 0.7828 (0.7805)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [81]  [2800/6250]  eta: 0:15:27  lr: 0.000011  grad: 0.2031 (0.3265)  loss: 0.7812 (0.7805)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [81]  [2900/6250]  eta: 0:15:00  lr: 0.000011  grad: 0.2087 (0.3316)  loss: 0.7778 (0.7804)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [81]  [3000/6250]  eta: 0:14:33  lr: 0.000011  grad: 0.1906 (0.3361)  loss: 0.7892 (0.7805)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [81]  [3100/6250]  eta: 0:14:06  lr: 0.000011  grad: 0.1936 (0.3337)  loss: 0.7805 (0.7806)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [81]  [3200/6250]  eta: 0:13:39  lr: 0.000011  grad: 0.1840 (0.3323)  loss: 0.7878 (0.7807)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [81]  [3300/6250]  eta: 0:13:12  lr: 0.000011  grad: 0.2005 (0.3299)  loss: 0.7847 (0.7808)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [81]  [3400/6250]  eta: 0:12:45  lr: 0.000011  grad: 0.1959 (0.3291)  loss: 0.7847 (0.7810)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [81]  [3500/6250]  eta: 0:12:18  lr: 0.000011  grad: 0.1923 (0.3361)  loss: 0.7912 (0.7811)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [81]  [3600/6250]  eta: 0:11:51  lr: 0.000011  grad: 0.1881 (0.3333)  loss: 0.7845 (0.7812)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [81]  [3700/6250]  eta: 0:11:24  lr: 0.000011  grad: 0.1816 (0.3324)  loss: 0.7815 (0.7813)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [81]  [3800/6250]  eta: 0:10:57  lr: 0.000011  grad: 0.1893 (0.3302)  loss: 0.7854 (0.7813)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [81]  [3900/6250]  eta: 0:10:30  lr: 0.000011  grad: 0.1936 (0.3354)  loss: 0.7848 (0.7814)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [81]  [4000/6250]  eta: 0:10:03  lr: 0.000011  grad: 0.1986 (0.3347)  loss: 0.7824 (0.7814)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [81]  [4100/6250]  eta: 0:09:36  lr: 0.000011  grad: 0.1897 (0.3361)  loss: 0.7822 (0.7815)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [81]  [4200/6250]  eta: 0:09:10  lr: 0.000011  grad: 0.1947 (0.3368)  loss: 0.7852 (0.7816)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [81]  [4300/6250]  eta: 0:08:43  lr: 0.000011  grad: 0.2054 (0.3382)  loss: 0.7809 (0.7817)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [81]  [4400/6250]  eta: 0:08:16  lr: 0.000011  grad: 0.2011 (0.3368)  loss: 0.7824 (0.7818)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [81]  [4500/6250]  eta: 0:07:49  lr: 0.000011  grad: 0.1886 (0.3379)  loss: 0.7833 (0.7818)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [81]  [4600/6250]  eta: 0:07:22  lr: 0.000011  grad: 0.1966 (0.3374)  loss: 0.7803 (0.7819)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [81]  [4700/6250]  eta: 0:06:55  lr: 0.000011  grad: 0.2086 (0.3403)  loss: 0.7803 (0.7818)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [81]  [4800/6250]  eta: 0:06:28  lr: 0.000011  grad: 0.1897 (0.3389)  loss: 0.7770 (0.7818)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [81]  [4900/6250]  eta: 0:06:01  lr: 0.000011  grad: 0.2030 (0.3405)  loss: 0.7802 (0.7817)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [81]  [5000/6250]  eta: 0:05:35  lr: 0.000011  grad: 0.1912 (0.3402)  loss: 0.7777 (0.7816)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [81]  [5100/6250]  eta: 0:05:08  lr: 0.000011  grad: 0.1957 (0.3423)  loss: 0.7758 (0.7816)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [81]  [5200/6250]  eta: 0:04:41  lr: 0.000011  grad: 0.2047 (0.3438)  loss: 0.7680 (0.7815)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [81]  [5300/6250]  eta: 0:04:14  lr: 0.000011  grad: 0.2037 (0.3473)  loss: 0.7681 (0.7814)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [81]  [5400/6250]  eta: 0:03:47  lr: 0.000011  grad: 0.1897 (0.3500)  loss: 0.7806 (0.7812)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [81]  [5500/6250]  eta: 0:03:20  lr: 0.000011  grad: 0.2099 (0.3491)  loss: 0.7734 (0.7811)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [81]  [5600/6250]  eta: 0:02:54  lr: 0.000011  grad: 0.1897 (0.3488)  loss: 0.7743 (0.7811)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [81]  [5700/6250]  eta: 0:02:27  lr: 0.000011  grad: 0.1978 (0.3480)  loss: 0.7715 (0.7810)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [81]  [5800/6250]  eta: 0:02:00  lr: 0.000011  grad: 0.1916 (0.3502)  loss: 0.7761 (0.7810)  time: 0.2703  data: 0.0002  max mem: 26157
Train: [81]  [5900/6250]  eta: 0:01:33  lr: 0.000011  grad: 0.2027 (0.3494)  loss: 0.7754 (0.7809)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [81]  [6000/6250]  eta: 0:01:06  lr: 0.000011  grad: 0.2094 (0.3493)  loss: 0.7729 (0.7807)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [81]  [6100/6250]  eta: 0:00:40  lr: 0.000011  grad: 0.1981 (0.3508)  loss: 0.7746 (0.7806)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [81]  [6200/6250]  eta: 0:00:13  lr: 0.000011  grad: 0.2201 (0.3527)  loss: 0.7661 (0.7805)  time: 0.2724  data: 0.0002  max mem: 26157
Train: [81]  [6249/6250]  eta: 0:00:00  lr: 0.000011  grad: 0.1923 (0.3523)  loss: 0.7753 (0.7804)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [81] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000011  grad: 0.1923 (0.3523)  loss: 0.7753 (0.7804)
Eval (hcp-train-subset): [81]  [ 0/62]  eta: 0:04:41  loss: 0.8159 (0.8159)  time: 4.5417  data: 4.4584  max mem: 26157
Eval (hcp-train-subset): [81]  [61/62]  eta: 0:00:00  loss: 0.7977 (0.7979)  time: 0.0890  data: 0.0066  max mem: 26157
Eval (hcp-train-subset): [81] Total time: 0:00:10 (0.1705 s / it)
Averaged stats (hcp-train-subset): loss: 0.7977 (0.7979)
Making plots (hcp-train-subset): example=38
Eval (hcp-val): [81]  [ 0/62]  eta: 0:03:33  loss: 0.8216 (0.8216)  time: 3.4404  data: 3.3241  max mem: 26157
Eval (hcp-val): [81]  [61/62]  eta: 0:00:00  loss: 0.8232 (0.8240)  time: 0.1039  data: 0.0215  max mem: 26157
Eval (hcp-val): [81] Total time: 0:00:11 (0.1887 s / it)
Averaged stats (hcp-val): loss: 0.8232 (0.8240)
Making plots (hcp-val): example=7
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [82]  [   0/6250]  eta: 6:16:27  lr: 0.000011  grad: 0.2058 (0.2058)  loss: 0.8298 (0.8298)  time: 3.6139  data: 3.2820  max mem: 26157
Train: [82]  [ 100/6250]  eta: 0:32:04  lr: 0.000011  grad: 0.2421 (0.4366)  loss: 0.7964 (0.7944)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [82]  [ 200/6250]  eta: 0:29:16  lr: 0.000011  grad: 0.2268 (0.3776)  loss: 0.7837 (0.7895)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [82]  [ 300/6250]  eta: 0:28:02  lr: 0.000011  grad: 0.2168 (0.3457)  loss: 0.7763 (0.7885)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [82]  [ 400/6250]  eta: 0:27:11  lr: 0.000011  grad: 0.2127 (0.3229)  loss: 0.7890 (0.7878)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [82]  [ 500/6250]  eta: 0:26:30  lr: 0.000011  grad: 0.2063 (0.3157)  loss: 0.7783 (0.7878)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [82]  [ 600/6250]  eta: 0:25:54  lr: 0.000011  grad: 0.1894 (0.3144)  loss: 0.7966 (0.7877)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [82]  [ 700/6250]  eta: 0:25:20  lr: 0.000011  grad: 0.1922 (0.3148)  loss: 0.7803 (0.7874)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [82]  [ 800/6250]  eta: 0:24:48  lr: 0.000011  grad: 0.2166 (0.3212)  loss: 0.7803 (0.7865)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [82]  [ 900/6250]  eta: 0:24:17  lr: 0.000011  grad: 0.1925 (0.3197)  loss: 0.7816 (0.7860)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [82]  [1000/6250]  eta: 0:23:47  lr: 0.000011  grad: 0.2051 (0.3371)  loss: 0.7749 (0.7857)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [82]  [1100/6250]  eta: 0:23:17  lr: 0.000011  grad: 0.1974 (0.3426)  loss: 0.7810 (0.7854)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [82]  [1200/6250]  eta: 0:22:48  lr: 0.000011  grad: 0.3991 (0.3687)  loss: 0.7874 (0.7852)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [82]  [1300/6250]  eta: 0:22:22  lr: 0.000011  grad: 0.1972 (0.3685)  loss: 0.7832 (0.7851)  time: 0.2658  data: 0.0001  max mem: 26157
Train: [82]  [1400/6250]  eta: 0:21:53  lr: 0.000010  grad: 0.1991 (0.3677)  loss: 0.7800 (0.7848)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [82]  [1500/6250]  eta: 0:21:25  lr: 0.000010  grad: 0.2075 (0.3614)  loss: 0.7771 (0.7844)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [82]  [1600/6250]  eta: 0:20:57  lr: 0.000010  grad: 0.1945 (0.3544)  loss: 0.7811 (0.7842)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [82]  [1700/6250]  eta: 0:20:29  lr: 0.000010  grad: 0.2113 (0.3543)  loss: 0.7809 (0.7839)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [82]  [1800/6250]  eta: 0:20:01  lr: 0.000010  grad: 0.1977 (0.3558)  loss: 0.7785 (0.7837)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [82]  [1900/6250]  eta: 0:19:33  lr: 0.000010  grad: 0.2064 (0.3638)  loss: 0.7811 (0.7835)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [82]  [2000/6250]  eta: 0:19:06  lr: 0.000010  grad: 0.2146 (0.3616)  loss: 0.7775 (0.7832)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [82]  [2100/6250]  eta: 0:18:38  lr: 0.000010  grad: 0.2131 (0.3600)  loss: 0.7777 (0.7830)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [82]  [2200/6250]  eta: 0:18:11  lr: 0.000010  grad: 0.2021 (0.3585)  loss: 0.7759 (0.7827)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [82]  [2300/6250]  eta: 0:17:44  lr: 0.000010  grad: 0.1938 (0.3547)  loss: 0.7733 (0.7824)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [82]  [2400/6250]  eta: 0:17:16  lr: 0.000010  grad: 0.1964 (0.3581)  loss: 0.7741 (0.7823)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [82]  [2500/6250]  eta: 0:16:49  lr: 0.000010  grad: 0.1915 (0.3591)  loss: 0.7819 (0.7822)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [82]  [2600/6250]  eta: 0:16:22  lr: 0.000010  grad: 0.2027 (0.3605)  loss: 0.7778 (0.7821)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [82]  [2700/6250]  eta: 0:15:55  lr: 0.000010  grad: 0.2136 (0.3613)  loss: 0.7707 (0.7820)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [82]  [2800/6250]  eta: 0:15:27  lr: 0.000010  grad: 0.1938 (0.3590)  loss: 0.7848 (0.7819)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [82]  [2900/6250]  eta: 0:15:00  lr: 0.000010  grad: 0.2039 (0.3590)  loss: 0.7748 (0.7818)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [82]  [3000/6250]  eta: 0:14:33  lr: 0.000010  grad: 0.2103 (0.3579)  loss: 0.7746 (0.7817)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [82]  [3100/6250]  eta: 0:14:06  lr: 0.000010  grad: 0.1926 (0.3564)  loss: 0.7796 (0.7815)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [82]  [3200/6250]  eta: 0:13:39  lr: 0.000010  grad: 0.1981 (0.3555)  loss: 0.7646 (0.7812)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [82]  [3300/6250]  eta: 0:13:12  lr: 0.000010  grad: 0.1926 (0.3574)  loss: 0.7764 (0.7810)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [82]  [3400/6250]  eta: 0:12:45  lr: 0.000010  grad: 0.1883 (0.3570)  loss: 0.7838 (0.7809)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [82]  [3500/6250]  eta: 0:12:18  lr: 0.000010  grad: 0.2296 (0.3580)  loss: 0.7722 (0.7807)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [82]  [3600/6250]  eta: 0:11:51  lr: 0.000010  grad: 0.1927 (0.3585)  loss: 0.7770 (0.7806)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [82]  [3700/6250]  eta: 0:11:24  lr: 0.000010  grad: 0.1986 (0.3583)  loss: 0.7801 (0.7806)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [82]  [3800/6250]  eta: 0:10:57  lr: 0.000010  grad: 0.1908 (0.3568)  loss: 0.7810 (0.7806)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [82]  [3900/6250]  eta: 0:10:30  lr: 0.000010  grad: 0.1931 (0.3556)  loss: 0.7844 (0.7805)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [82]  [4000/6250]  eta: 0:10:03  lr: 0.000010  grad: 0.2066 (0.3556)  loss: 0.7732 (0.7804)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [82]  [4100/6250]  eta: 0:09:36  lr: 0.000010  grad: 0.2049 (0.3583)  loss: 0.7806 (0.7803)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [82]  [4200/6250]  eta: 0:09:10  lr: 0.000010  grad: 0.2117 (0.3599)  loss: 0.7771 (0.7803)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [82]  [4300/6250]  eta: 0:08:43  lr: 0.000010  grad: 0.2161 (0.3572)  loss: 0.7719 (0.7802)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [82]  [4400/6250]  eta: 0:08:16  lr: 0.000010  grad: 0.2136 (0.3589)  loss: 0.7713 (0.7801)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [82]  [4500/6250]  eta: 0:07:49  lr: 0.000010  grad: 0.1970 (0.3595)  loss: 0.7783 (0.7800)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [82]  [4600/6250]  eta: 0:07:22  lr: 0.000010  grad: 0.1979 (0.3588)  loss: 0.7721 (0.7800)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [82]  [4700/6250]  eta: 0:06:55  lr: 0.000010  grad: 0.2061 (0.3586)  loss: 0.7728 (0.7799)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [82]  [4800/6250]  eta: 0:06:28  lr: 0.000010  grad: 0.2032 (0.3612)  loss: 0.7748 (0.7798)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [82]  [4900/6250]  eta: 0:06:01  lr: 0.000010  grad: 0.2083 (0.3641)  loss: 0.7856 (0.7798)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [82]  [5000/6250]  eta: 0:05:35  lr: 0.000010  grad: 0.2074 (0.3684)  loss: 0.7794 (0.7798)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [82]  [5100/6250]  eta: 0:05:08  lr: 0.000010  grad: 0.2043 (0.3694)  loss: 0.7790 (0.7798)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [82]  [5200/6250]  eta: 0:04:41  lr: 0.000010  grad: 0.2129 (0.3691)  loss: 0.7766 (0.7797)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [82]  [5300/6250]  eta: 0:04:14  lr: 0.000010  grad: 0.2010 (0.3708)  loss: 0.7796 (0.7797)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [82]  [5400/6250]  eta: 0:03:47  lr: 0.000010  grad: 0.2019 (0.3712)  loss: 0.7763 (0.7797)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [82]  [5500/6250]  eta: 0:03:20  lr: 0.000010  grad: 0.2040 (0.3725)  loss: 0.7838 (0.7797)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [82]  [5600/6250]  eta: 0:02:54  lr: 0.000010  grad: 0.2233 (0.3753)  loss: 0.7795 (0.7796)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [82]  [5700/6250]  eta: 0:02:27  lr: 0.000010  grad: 0.2026 (0.3795)  loss: 0.7767 (0.7797)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [82]  [5800/6250]  eta: 0:02:00  lr: 0.000010  grad: 0.1896 (0.3795)  loss: 0.7897 (0.7798)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [82]  [5900/6250]  eta: 0:01:33  lr: 0.000010  grad: 0.1951 (0.3792)  loss: 0.7726 (0.7799)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [82]  [6000/6250]  eta: 0:01:06  lr: 0.000010  grad: 0.1995 (0.3824)  loss: 0.7804 (0.7799)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [82]  [6100/6250]  eta: 0:00:40  lr: 0.000010  grad: 0.2009 (0.3838)  loss: 0.7825 (0.7799)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [82]  [6200/6250]  eta: 0:00:13  lr: 0.000010  grad: 0.2084 (0.3817)  loss: 0.7749 (0.7799)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [82]  [6249/6250]  eta: 0:00:00  lr: 0.000010  grad: 0.1986 (0.3804)  loss: 0.7811 (0.7800)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [82] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000010  grad: 0.1986 (0.3804)  loss: 0.7811 (0.7800)
Eval (hcp-train-subset): [82]  [ 0/62]  eta: 0:03:24  loss: 0.8077 (0.8077)  time: 3.2951  data: 3.2024  max mem: 26157
Eval (hcp-train-subset): [82]  [61/62]  eta: 0:00:00  loss: 0.7924 (0.7955)  time: 0.0946  data: 0.0124  max mem: 26157
Eval (hcp-train-subset): [82] Total time: 0:00:10 (0.1643 s / it)
Averaged stats (hcp-train-subset): loss: 0.7924 (0.7955)
Making plots (hcp-train-subset): example=25
Eval (hcp-val): [82]  [ 0/62]  eta: 0:04:14  loss: 0.8175 (0.8175)  time: 4.1091  data: 4.0270  max mem: 26157
Eval (hcp-val): [82]  [61/62]  eta: 0:00:00  loss: 0.8226 (0.8235)  time: 0.0824  data: 0.0001  max mem: 26157
Eval (hcp-val): [82] Total time: 0:00:10 (0.1631 s / it)
Averaged stats (hcp-val): loss: 0.8226 (0.8235)
Making plots (hcp-val): example=38
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [83]  [   0/6250]  eta: 6:47:59  lr: 0.000010  grad: 0.5127 (0.5127)  loss: 0.8106 (0.8106)  time: 3.9167  data: 3.5920  max mem: 26157
Train: [83]  [ 100/6250]  eta: 0:32:00  lr: 0.000010  grad: 0.2721 (0.4467)  loss: 0.7994 (0.7963)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [83]  [ 200/6250]  eta: 0:29:12  lr: 0.000010  grad: 0.2110 (0.4739)  loss: 0.7882 (0.7895)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [83]  [ 300/6250]  eta: 0:27:58  lr: 0.000010  grad: 0.2374 (0.4408)  loss: 0.7802 (0.7863)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [83]  [ 400/6250]  eta: 0:27:09  lr: 0.000010  grad: 0.2450 (0.4961)  loss: 0.7778 (0.7840)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [83]  [ 500/6250]  eta: 0:26:28  lr: 0.000010  grad: 0.2317 (0.5609)  loss: 0.7803 (0.7822)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [83]  [ 600/6250]  eta: 0:25:52  lr: 0.000010  grad: 0.2644 (0.5677)  loss: 0.7803 (0.7816)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [83]  [ 700/6250]  eta: 0:25:18  lr: 0.000009  grad: 0.2221 (0.5869)  loss: 0.7755 (0.7811)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [83]  [ 800/6250]  eta: 0:24:47  lr: 0.000009  grad: 0.2129 (0.5759)  loss: 0.7772 (0.7810)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [83]  [ 900/6250]  eta: 0:24:16  lr: 0.000009  grad: 0.2027 (0.5566)  loss: 0.7825 (0.7808)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [83]  [1000/6250]  eta: 0:23:46  lr: 0.000009  grad: 0.2064 (0.5498)  loss: 0.7889 (0.7807)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [83]  [1100/6250]  eta: 0:23:17  lr: 0.000009  grad: 0.2173 (0.5335)  loss: 0.7805 (0.7807)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [83]  [1200/6250]  eta: 0:22:48  lr: 0.000009  grad: 0.1935 (0.5155)  loss: 0.7866 (0.7808)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [83]  [1300/6250]  eta: 0:22:19  lr: 0.000009  grad: 0.1943 (0.5053)  loss: 0.7772 (0.7807)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [83]  [1400/6250]  eta: 0:21:51  lr: 0.000009  grad: 0.1847 (0.4997)  loss: 0.7930 (0.7809)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [83]  [1500/6250]  eta: 0:21:23  lr: 0.000009  grad: 0.1968 (0.4994)  loss: 0.7771 (0.7809)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [83]  [1600/6250]  eta: 0:20:55  lr: 0.000009  grad: 0.1931 (0.4835)  loss: 0.7726 (0.7809)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [83]  [1700/6250]  eta: 0:20:27  lr: 0.000009  grad: 0.2014 (0.4781)  loss: 0.7786 (0.7809)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [83]  [1800/6250]  eta: 0:19:59  lr: 0.000009  grad: 0.2014 (0.4711)  loss: 0.7795 (0.7808)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [83]  [1900/6250]  eta: 0:19:32  lr: 0.000009  grad: 0.1905 (0.4633)  loss: 0.7797 (0.7809)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [83]  [2000/6250]  eta: 0:19:04  lr: 0.000009  grad: 0.1882 (0.4533)  loss: 0.7741 (0.7809)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [83]  [2100/6250]  eta: 0:18:37  lr: 0.000009  grad: 0.1863 (0.4508)  loss: 0.7874 (0.7810)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [83]  [2200/6250]  eta: 0:18:10  lr: 0.000009  grad: 0.1921 (0.4447)  loss: 0.7755 (0.7810)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [83]  [2300/6250]  eta: 0:17:42  lr: 0.000009  grad: 0.1935 (0.4377)  loss: 0.7803 (0.7810)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [83]  [2400/6250]  eta: 0:17:15  lr: 0.000009  grad: 0.1935 (0.4323)  loss: 0.7751 (0.7811)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [83]  [2500/6250]  eta: 0:16:48  lr: 0.000009  grad: 0.1771 (0.4308)  loss: 0.7843 (0.7812)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [83]  [2600/6250]  eta: 0:16:21  lr: 0.000009  grad: 0.1949 (0.4247)  loss: 0.7771 (0.7812)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [83]  [2700/6250]  eta: 0:15:54  lr: 0.000009  grad: 0.1873 (0.4226)  loss: 0.7877 (0.7812)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [83]  [2800/6250]  eta: 0:15:26  lr: 0.000009  grad: 0.1863 (0.4175)  loss: 0.7860 (0.7814)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [83]  [2900/6250]  eta: 0:14:59  lr: 0.000009  grad: 0.1882 (0.4124)  loss: 0.7821 (0.7815)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [83]  [3000/6250]  eta: 0:14:32  lr: 0.000009  grad: 0.2060 (0.4118)  loss: 0.7788 (0.7816)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [83]  [3100/6250]  eta: 0:14:05  lr: 0.000009  grad: 0.2187 (0.4097)  loss: 0.7746 (0.7816)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [83]  [3200/6250]  eta: 0:13:38  lr: 0.000009  grad: 0.1979 (0.4048)  loss: 0.7813 (0.7815)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [83]  [3300/6250]  eta: 0:13:11  lr: 0.000009  grad: 0.1962 (0.4093)  loss: 0.7882 (0.7815)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [83]  [3400/6250]  eta: 0:12:44  lr: 0.000009  grad: 0.2040 (0.4074)  loss: 0.7885 (0.7815)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [83]  [3500/6250]  eta: 0:12:17  lr: 0.000009  grad: 0.2063 (0.4059)  loss: 0.7848 (0.7815)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [83]  [3600/6250]  eta: 0:11:51  lr: 0.000009  grad: 0.1987 (0.4079)  loss: 0.7779 (0.7814)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [83]  [3700/6250]  eta: 0:11:24  lr: 0.000009  grad: 0.2000 (0.4102)  loss: 0.7825 (0.7815)  time: 0.2657  data: 0.0001  max mem: 26157
Train: [83]  [3800/6250]  eta: 0:10:57  lr: 0.000009  grad: 0.1999 (0.4066)  loss: 0.7836 (0.7815)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [83]  [3900/6250]  eta: 0:10:30  lr: 0.000009  grad: 0.2105 (0.4039)  loss: 0.7768 (0.7814)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [83]  [4000/6250]  eta: 0:10:03  lr: 0.000009  grad: 0.2135 (0.4037)  loss: 0.7792 (0.7814)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [83]  [4100/6250]  eta: 0:09:36  lr: 0.000009  grad: 0.2030 (0.4007)  loss: 0.7793 (0.7814)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [83]  [4200/6250]  eta: 0:09:09  lr: 0.000009  grad: 0.1916 (0.3983)  loss: 0.7848 (0.7814)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [83]  [4300/6250]  eta: 0:08:42  lr: 0.000009  grad: 0.2030 (0.3989)  loss: 0.7810 (0.7815)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [83]  [4400/6250]  eta: 0:08:15  lr: 0.000009  grad: 0.1927 (0.4009)  loss: 0.7963 (0.7817)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [83]  [4500/6250]  eta: 0:07:49  lr: 0.000009  grad: 0.1952 (0.3979)  loss: 0.7805 (0.7818)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [83]  [4600/6250]  eta: 0:07:22  lr: 0.000009  grad: 0.1973 (0.3989)  loss: 0.7822 (0.7819)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [83]  [4700/6250]  eta: 0:06:55  lr: 0.000009  grad: 0.2004 (0.3968)  loss: 0.7864 (0.7820)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [83]  [4800/6250]  eta: 0:06:28  lr: 0.000009  grad: 0.1938 (0.4002)  loss: 0.7824 (0.7821)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [83]  [4900/6250]  eta: 0:06:01  lr: 0.000009  grad: 0.1996 (0.3983)  loss: 0.7769 (0.7821)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [83]  [5000/6250]  eta: 0:05:34  lr: 0.000009  grad: 0.2025 (0.3959)  loss: 0.7859 (0.7822)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [83]  [5100/6250]  eta: 0:05:08  lr: 0.000009  grad: 0.1956 (0.3960)  loss: 0.7824 (0.7822)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [83]  [5200/6250]  eta: 0:04:41  lr: 0.000009  grad: 0.1931 (0.3931)  loss: 0.7897 (0.7822)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [83]  [5300/6250]  eta: 0:04:14  lr: 0.000009  grad: 0.1997 (0.3917)  loss: 0.7770 (0.7823)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [83]  [5400/6250]  eta: 0:03:47  lr: 0.000009  grad: 0.2009 (0.3917)  loss: 0.7781 (0.7823)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [83]  [5500/6250]  eta: 0:03:20  lr: 0.000009  grad: 0.1958 (0.3909)  loss: 0.7742 (0.7824)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [83]  [5600/6250]  eta: 0:02:54  lr: 0.000009  grad: 0.1978 (0.3905)  loss: 0.7859 (0.7824)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [83]  [5700/6250]  eta: 0:02:27  lr: 0.000009  grad: 0.2064 (0.3928)  loss: 0.7752 (0.7824)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [83]  [5800/6250]  eta: 0:02:00  lr: 0.000009  grad: 0.2177 (0.3942)  loss: 0.7750 (0.7824)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [83]  [5900/6250]  eta: 0:01:33  lr: 0.000009  grad: 0.1834 (0.3935)  loss: 0.7849 (0.7824)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [83]  [6000/6250]  eta: 0:01:06  lr: 0.000009  grad: 0.2142 (0.3942)  loss: 0.7805 (0.7823)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [83]  [6100/6250]  eta: 0:00:40  lr: 0.000009  grad: 0.2172 (0.3956)  loss: 0.7787 (0.7823)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [83]  [6200/6250]  eta: 0:00:13  lr: 0.000009  grad: 0.2478 (0.3957)  loss: 0.7846 (0.7823)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [83]  [6249/6250]  eta: 0:00:00  lr: 0.000009  grad: 0.2068 (0.3953)  loss: 0.7829 (0.7823)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [83] Total time: 0:27:56 (0.2683 s / it)
Averaged stats: lr: 0.000009  grad: 0.2068 (0.3953)  loss: 0.7829 (0.7823)
Eval (hcp-train-subset): [83]  [ 0/62]  eta: 0:04:05  loss: 0.8027 (0.8027)  time: 3.9616  data: 3.8536  max mem: 26157
Eval (hcp-train-subset): [83]  [61/62]  eta: 0:00:00  loss: 0.7917 (0.7953)  time: 0.1035  data: 0.0193  max mem: 26157
Eval (hcp-train-subset): [83] Total time: 0:00:11 (0.1864 s / it)
Averaged stats (hcp-train-subset): loss: 0.7917 (0.7953)
Making plots (hcp-train-subset): example=60
Eval (hcp-val): [83]  [ 0/62]  eta: 0:04:18  loss: 0.8219 (0.8219)  time: 4.1754  data: 4.0922  max mem: 26157
Eval (hcp-val): [83]  [61/62]  eta: 0:00:00  loss: 0.8229 (0.8234)  time: 0.0945  data: 0.0104  max mem: 26157
Eval (hcp-val): [83] Total time: 0:00:10 (0.1764 s / it)
Averaged stats (hcp-val): loss: 0.8229 (0.8234)
Making plots (hcp-val): example=61
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [84]  [   0/6250]  eta: 6:21:02  lr: 0.000009  grad: 0.2163 (0.2163)  loss: 0.8301 (0.8301)  time: 3.6579  data: 3.3164  max mem: 26157
Train: [84]  [ 100/6250]  eta: 0:31:12  lr: 0.000009  grad: 0.2311 (0.3741)  loss: 0.7980 (0.7987)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [84]  [ 200/6250]  eta: 0:28:48  lr: 0.000009  grad: 0.2152 (0.3536)  loss: 0.7914 (0.7960)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [84]  [ 300/6250]  eta: 0:27:44  lr: 0.000008  grad: 0.2042 (0.3359)  loss: 0.7872 (0.7929)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [84]  [ 400/6250]  eta: 0:26:58  lr: 0.000008  grad: 0.2228 (0.3226)  loss: 0.7827 (0.7900)  time: 0.2689  data: 0.0002  max mem: 26157
Train: [84]  [ 500/6250]  eta: 0:26:19  lr: 0.000008  grad: 0.2086 (0.3301)  loss: 0.7799 (0.7887)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [84]  [ 600/6250]  eta: 0:25:45  lr: 0.000008  grad: 0.2047 (0.3213)  loss: 0.7823 (0.7879)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [ 700/6250]  eta: 0:25:12  lr: 0.000008  grad: 0.2045 (0.3224)  loss: 0.7860 (0.7869)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [ 800/6250]  eta: 0:24:41  lr: 0.000008  grad: 0.1934 (0.3224)  loss: 0.7795 (0.7860)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [ 900/6250]  eta: 0:24:11  lr: 0.000008  grad: 0.1974 (0.3173)  loss: 0.7927 (0.7858)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [84]  [1000/6250]  eta: 0:23:42  lr: 0.000008  grad: 0.1907 (0.3298)  loss: 0.7812 (0.7856)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [84]  [1100/6250]  eta: 0:23:13  lr: 0.000008  grad: 0.1985 (0.3281)  loss: 0.7866 (0.7854)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [1200/6250]  eta: 0:22:45  lr: 0.000008  grad: 0.2066 (0.3302)  loss: 0.7834 (0.7852)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [1300/6250]  eta: 0:22:16  lr: 0.000008  grad: 0.1979 (0.3218)  loss: 0.7786 (0.7850)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [84]  [1400/6250]  eta: 0:21:48  lr: 0.000008  grad: 0.1780 (0.3200)  loss: 0.7836 (0.7847)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [84]  [1500/6250]  eta: 0:21:20  lr: 0.000008  grad: 0.2043 (0.3203)  loss: 0.7805 (0.7844)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [84]  [1600/6250]  eta: 0:20:52  lr: 0.000008  grad: 0.1990 (0.3147)  loss: 0.7840 (0.7843)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [84]  [1700/6250]  eta: 0:20:25  lr: 0.000008  grad: 0.1908 (0.3190)  loss: 0.7823 (0.7839)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [84]  [1800/6250]  eta: 0:19:57  lr: 0.000008  grad: 0.1913 (0.3151)  loss: 0.7835 (0.7837)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [84]  [1900/6250]  eta: 0:19:30  lr: 0.000008  grad: 0.1985 (0.3257)  loss: 0.7712 (0.7835)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [2000/6250]  eta: 0:19:03  lr: 0.000008  grad: 0.1954 (0.3266)  loss: 0.7847 (0.7832)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [84]  [2100/6250]  eta: 0:18:35  lr: 0.000008  grad: 0.1939 (0.3242)  loss: 0.7866 (0.7832)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [84]  [2200/6250]  eta: 0:18:08  lr: 0.000008  grad: 0.2041 (0.3270)  loss: 0.7794 (0.7830)  time: 0.2685  data: 0.0001  max mem: 26157
Train: [84]  [2300/6250]  eta: 0:17:41  lr: 0.000008  grad: 0.1903 (0.3238)  loss: 0.7789 (0.7830)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [2400/6250]  eta: 0:17:14  lr: 0.000008  grad: 0.2063 (0.3260)  loss: 0.7820 (0.7830)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [84]  [2500/6250]  eta: 0:16:47  lr: 0.000008  grad: 0.1860 (0.3242)  loss: 0.7843 (0.7831)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [2600/6250]  eta: 0:16:20  lr: 0.000008  grad: 0.1926 (0.3242)  loss: 0.7723 (0.7831)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [84]  [2700/6250]  eta: 0:15:53  lr: 0.000008  grad: 0.1946 (0.3266)  loss: 0.7840 (0.7830)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [84]  [2800/6250]  eta: 0:15:26  lr: 0.000008  grad: 0.1878 (0.3261)  loss: 0.7855 (0.7829)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [84]  [2900/6250]  eta: 0:14:59  lr: 0.000008  grad: 0.1850 (0.3273)  loss: 0.7843 (0.7829)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [3000/6250]  eta: 0:14:31  lr: 0.000008  grad: 0.2030 (0.3279)  loss: 0.7805 (0.7829)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [84]  [3100/6250]  eta: 0:14:05  lr: 0.000008  grad: 0.1963 (0.3278)  loss: 0.7839 (0.7828)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [84]  [3200/6250]  eta: 0:13:38  lr: 0.000008  grad: 0.2055 (0.3325)  loss: 0.7830 (0.7828)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [84]  [3300/6250]  eta: 0:13:11  lr: 0.000008  grad: 0.2103 (0.3306)  loss: 0.7813 (0.7827)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [84]  [3400/6250]  eta: 0:12:44  lr: 0.000008  grad: 0.2036 (0.3287)  loss: 0.7782 (0.7829)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [84]  [3500/6250]  eta: 0:12:17  lr: 0.000008  grad: 0.1969 (0.3293)  loss: 0.7846 (0.7829)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [84]  [3600/6250]  eta: 0:11:50  lr: 0.000008  grad: 0.2013 (0.3272)  loss: 0.7871 (0.7830)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [84]  [3700/6250]  eta: 0:11:23  lr: 0.000008  grad: 0.1897 (0.3274)  loss: 0.7860 (0.7830)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [3800/6250]  eta: 0:10:56  lr: 0.000008  grad: 0.1923 (0.3263)  loss: 0.7846 (0.7831)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [84]  [3900/6250]  eta: 0:10:29  lr: 0.000008  grad: 0.1991 (0.3266)  loss: 0.7859 (0.7831)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [84]  [4000/6250]  eta: 0:10:02  lr: 0.000008  grad: 0.2107 (0.3275)  loss: 0.7836 (0.7832)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [84]  [4100/6250]  eta: 0:09:36  lr: 0.000008  grad: 0.2058 (0.3274)  loss: 0.7889 (0.7832)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [4200/6250]  eta: 0:09:09  lr: 0.000008  grad: 0.1972 (0.3300)  loss: 0.7857 (0.7831)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [84]  [4300/6250]  eta: 0:08:42  lr: 0.000008  grad: 0.1926 (0.3331)  loss: 0.7840 (0.7832)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [4400/6250]  eta: 0:08:15  lr: 0.000008  grad: 0.1979 (0.3331)  loss: 0.7897 (0.7833)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [84]  [4500/6250]  eta: 0:07:48  lr: 0.000008  grad: 0.2032 (0.3314)  loss: 0.7869 (0.7833)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [84]  [4600/6250]  eta: 0:07:21  lr: 0.000008  grad: 0.2020 (0.3327)  loss: 0.7797 (0.7833)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [84]  [4700/6250]  eta: 0:06:55  lr: 0.000008  grad: 0.2075 (0.3341)  loss: 0.7819 (0.7832)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [84]  [4800/6250]  eta: 0:06:28  lr: 0.000008  grad: 0.2220 (0.3371)  loss: 0.7746 (0.7830)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [84]  [4900/6250]  eta: 0:06:01  lr: 0.000008  grad: 0.2100 (0.3400)  loss: 0.7754 (0.7829)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [84]  [5000/6250]  eta: 0:05:34  lr: 0.000008  grad: 0.2097 (0.3411)  loss: 0.7702 (0.7828)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [5100/6250]  eta: 0:05:07  lr: 0.000008  grad: 0.2189 (0.3429)  loss: 0.7703 (0.7827)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [84]  [5200/6250]  eta: 0:04:41  lr: 0.000008  grad: 0.2251 (0.3414)  loss: 0.7839 (0.7826)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [84]  [5300/6250]  eta: 0:04:14  lr: 0.000008  grad: 0.2086 (0.3409)  loss: 0.7821 (0.7825)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [84]  [5400/6250]  eta: 0:03:47  lr: 0.000008  grad: 0.2177 (0.3411)  loss: 0.7770 (0.7825)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [5500/6250]  eta: 0:03:20  lr: 0.000008  grad: 0.2089 (0.3426)  loss: 0.7775 (0.7825)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [84]  [5600/6250]  eta: 0:02:54  lr: 0.000008  grad: 0.2315 (0.3452)  loss: 0.7826 (0.7824)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [84]  [5700/6250]  eta: 0:02:27  lr: 0.000008  grad: 0.1960 (0.3437)  loss: 0.7857 (0.7824)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [84]  [5800/6250]  eta: 0:02:00  lr: 0.000008  grad: 0.2074 (0.3435)  loss: 0.7798 (0.7824)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [84]  [5900/6250]  eta: 0:01:33  lr: 0.000008  grad: 0.2014 (0.3429)  loss: 0.7744 (0.7824)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [84]  [6000/6250]  eta: 0:01:06  lr: 0.000008  grad: 0.2098 (0.3436)  loss: 0.7679 (0.7824)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [6100/6250]  eta: 0:00:40  lr: 0.000008  grad: 0.2105 (0.3456)  loss: 0.7832 (0.7823)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [84]  [6200/6250]  eta: 0:00:13  lr: 0.000008  grad: 0.2119 (0.3460)  loss: 0.7821 (0.7823)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [84]  [6249/6250]  eta: 0:00:00  lr: 0.000008  grad: 0.2009 (0.3451)  loss: 0.7861 (0.7823)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [84] Total time: 0:27:55 (0.2681 s / it)
Averaged stats: lr: 0.000008  grad: 0.2009 (0.3451)  loss: 0.7861 (0.7823)
Eval (hcp-train-subset): [84]  [ 0/62]  eta: 0:03:17  loss: 0.8040 (0.8040)  time: 3.1897  data: 3.0815  max mem: 26157
Eval (hcp-train-subset): [84]  [61/62]  eta: 0:00:00  loss: 0.7893 (0.7938)  time: 0.0844  data: 0.0001  max mem: 26157
Eval (hcp-train-subset): [84] Total time: 0:00:10 (0.1695 s / it)
Averaged stats (hcp-train-subset): loss: 0.7893 (0.7938)
Making plots (hcp-train-subset): example=56
Eval (hcp-val): [84]  [ 0/62]  eta: 0:04:40  loss: 0.8131 (0.8131)  time: 4.5228  data: 4.4400  max mem: 26157
Eval (hcp-val): [84]  [61/62]  eta: 0:00:00  loss: 0.8215 (0.8232)  time: 0.0820  data: 0.0001  max mem: 26157
Eval (hcp-val): [84] Total time: 0:00:10 (0.1697 s / it)
Averaged stats (hcp-val): loss: 0.8215 (0.8232)
Making plots (hcp-val): example=2
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [85]  [   0/6250]  eta: 5:53:53  lr: 0.000008  grad: 0.2158 (0.2158)  loss: 0.8463 (0.8463)  time: 3.3974  data: 3.0715  max mem: 26157
Train: [85]  [ 100/6250]  eta: 0:32:20  lr: 0.000008  grad: 0.2053 (0.4266)  loss: 0.7835 (0.7896)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [85]  [ 200/6250]  eta: 0:29:22  lr: 0.000008  grad: 0.2322 (0.3920)  loss: 0.7797 (0.7861)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [ 300/6250]  eta: 0:28:04  lr: 0.000007  grad: 0.2496 (0.4125)  loss: 0.7863 (0.7853)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [85]  [ 400/6250]  eta: 0:27:12  lr: 0.000007  grad: 0.2021 (0.3939)  loss: 0.7883 (0.7850)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [85]  [ 500/6250]  eta: 0:26:31  lr: 0.000007  grad: 0.2162 (0.4093)  loss: 0.7729 (0.7851)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [85]  [ 600/6250]  eta: 0:25:55  lr: 0.000007  grad: 0.2035 (0.4087)  loss: 0.7823 (0.7853)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [85]  [ 700/6250]  eta: 0:25:20  lr: 0.000007  grad: 0.2020 (0.4104)  loss: 0.7811 (0.7853)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [85]  [ 800/6250]  eta: 0:24:49  lr: 0.000007  grad: 0.1899 (0.4022)  loss: 0.7730 (0.7849)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [85]  [ 900/6250]  eta: 0:24:19  lr: 0.000007  grad: 0.2093 (0.3853)  loss: 0.7800 (0.7844)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [85]  [1000/6250]  eta: 0:23:48  lr: 0.000007  grad: 0.2115 (0.3909)  loss: 0.7811 (0.7842)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [85]  [1100/6250]  eta: 0:23:18  lr: 0.000007  grad: 0.1915 (0.3858)  loss: 0.7843 (0.7840)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [1200/6250]  eta: 0:22:49  lr: 0.000007  grad: 0.2067 (0.3740)  loss: 0.7780 (0.7836)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [1300/6250]  eta: 0:22:21  lr: 0.000007  grad: 0.2191 (0.3791)  loss: 0.7746 (0.7834)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [85]  [1400/6250]  eta: 0:21:52  lr: 0.000007  grad: 0.2411 (0.3720)  loss: 0.7795 (0.7831)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [1500/6250]  eta: 0:21:24  lr: 0.000007  grad: 0.2021 (0.3670)  loss: 0.7884 (0.7828)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [85]  [1600/6250]  eta: 0:20:56  lr: 0.000007  grad: 0.2005 (0.3645)  loss: 0.7794 (0.7828)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [85]  [1700/6250]  eta: 0:20:28  lr: 0.000007  grad: 0.2005 (0.3651)  loss: 0.7781 (0.7827)  time: 0.2683  data: 0.0002  max mem: 26157
Train: [85]  [1800/6250]  eta: 0:20:01  lr: 0.000007  grad: 0.2066 (0.3647)  loss: 0.7842 (0.7827)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [85]  [1900/6250]  eta: 0:19:33  lr: 0.000007  grad: 0.2086 (0.3695)  loss: 0.7829 (0.7826)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [85]  [2000/6250]  eta: 0:19:05  lr: 0.000007  grad: 0.1985 (0.3687)  loss: 0.7728 (0.7824)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [85]  [2100/6250]  eta: 0:18:38  lr: 0.000007  grad: 0.1975 (0.3658)  loss: 0.7782 (0.7823)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [85]  [2200/6250]  eta: 0:18:10  lr: 0.000007  grad: 0.1974 (0.3631)  loss: 0.7812 (0.7823)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [85]  [2300/6250]  eta: 0:17:43  lr: 0.000007  grad: 0.1967 (0.3613)  loss: 0.7855 (0.7823)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [2400/6250]  eta: 0:17:16  lr: 0.000007  grad: 0.2018 (0.3590)  loss: 0.7834 (0.7823)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [85]  [2500/6250]  eta: 0:16:48  lr: 0.000007  grad: 0.2131 (0.3632)  loss: 0.7768 (0.7821)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [85]  [2600/6250]  eta: 0:16:21  lr: 0.000007  grad: 0.1950 (0.3610)  loss: 0.7703 (0.7820)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [85]  [2700/6250]  eta: 0:15:54  lr: 0.000007  grad: 0.2061 (0.3590)  loss: 0.7746 (0.7818)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [85]  [2800/6250]  eta: 0:15:27  lr: 0.000007  grad: 0.2152 (0.3588)  loss: 0.7811 (0.7818)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [85]  [2900/6250]  eta: 0:15:00  lr: 0.000007  grad: 0.2111 (0.3633)  loss: 0.7777 (0.7817)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [3000/6250]  eta: 0:14:33  lr: 0.000007  grad: 0.2056 (0.3639)  loss: 0.7782 (0.7816)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [85]  [3100/6250]  eta: 0:14:06  lr: 0.000007  grad: 0.2011 (0.3651)  loss: 0.7810 (0.7815)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [85]  [3200/6250]  eta: 0:13:39  lr: 0.000007  grad: 0.2015 (0.3627)  loss: 0.7738 (0.7814)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [85]  [3300/6250]  eta: 0:13:12  lr: 0.000007  grad: 0.2057 (0.3618)  loss: 0.7803 (0.7812)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [85]  [3400/6250]  eta: 0:12:45  lr: 0.000007  grad: 0.1961 (0.3607)  loss: 0.7812 (0.7811)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [85]  [3500/6250]  eta: 0:12:18  lr: 0.000007  grad: 0.1911 (0.3578)  loss: 0.7818 (0.7811)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [85]  [3600/6250]  eta: 0:11:51  lr: 0.000007  grad: 0.2135 (0.3547)  loss: 0.7803 (0.7811)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [85]  [3700/6250]  eta: 0:11:24  lr: 0.000007  grad: 0.2192 (0.3558)  loss: 0.7820 (0.7812)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [85]  [3800/6250]  eta: 0:10:57  lr: 0.000007  grad: 0.2175 (0.3557)  loss: 0.7752 (0.7812)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [85]  [3900/6250]  eta: 0:10:30  lr: 0.000007  grad: 0.1916 (0.3582)  loss: 0.7771 (0.7813)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [85]  [4000/6250]  eta: 0:10:03  lr: 0.000007  grad: 0.2054 (0.3609)  loss: 0.7787 (0.7812)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [4100/6250]  eta: 0:09:37  lr: 0.000007  grad: 0.2070 (0.3598)  loss: 0.7820 (0.7812)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [85]  [4200/6250]  eta: 0:09:10  lr: 0.000007  grad: 0.2131 (0.3575)  loss: 0.7780 (0.7812)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [85]  [4300/6250]  eta: 0:08:43  lr: 0.000007  grad: 0.2127 (0.3561)  loss: 0.7851 (0.7812)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [85]  [4400/6250]  eta: 0:08:16  lr: 0.000007  grad: 0.1989 (0.3563)  loss: 0.7830 (0.7812)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [85]  [4500/6250]  eta: 0:07:49  lr: 0.000007  grad: 0.1935 (0.3556)  loss: 0.7896 (0.7813)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [85]  [4600/6250]  eta: 0:07:22  lr: 0.000007  grad: 0.1941 (0.3539)  loss: 0.7810 (0.7813)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [85]  [4700/6250]  eta: 0:06:55  lr: 0.000007  grad: 0.1933 (0.3547)  loss: 0.7814 (0.7812)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [85]  [4800/6250]  eta: 0:06:28  lr: 0.000007  grad: 0.2095 (0.3557)  loss: 0.7772 (0.7812)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [85]  [4900/6250]  eta: 0:06:02  lr: 0.000007  grad: 0.2046 (0.3555)  loss: 0.7858 (0.7813)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [85]  [5000/6250]  eta: 0:05:35  lr: 0.000007  grad: 0.2032 (0.3576)  loss: 0.7831 (0.7813)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [85]  [5100/6250]  eta: 0:05:08  lr: 0.000007  grad: 0.2027 (0.3614)  loss: 0.7897 (0.7814)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [85]  [5200/6250]  eta: 0:04:41  lr: 0.000007  grad: 0.1989 (0.3620)  loss: 0.7866 (0.7814)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [85]  [5300/6250]  eta: 0:04:14  lr: 0.000007  grad: 0.2019 (0.3645)  loss: 0.7822 (0.7814)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [85]  [5400/6250]  eta: 0:03:47  lr: 0.000007  grad: 0.2082 (0.3643)  loss: 0.7867 (0.7814)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [85]  [5500/6250]  eta: 0:03:21  lr: 0.000007  grad: 0.2040 (0.3646)  loss: 0.7890 (0.7815)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [85]  [5600/6250]  eta: 0:02:54  lr: 0.000007  grad: 0.2122 (0.3630)  loss: 0.7761 (0.7815)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [5700/6250]  eta: 0:02:27  lr: 0.000007  grad: 0.2006 (0.3613)  loss: 0.7822 (0.7816)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [85]  [5800/6250]  eta: 0:02:00  lr: 0.000007  grad: 0.1956 (0.3666)  loss: 0.7814 (0.7816)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [85]  [5900/6250]  eta: 0:01:33  lr: 0.000007  grad: 0.1992 (0.3652)  loss: 0.7798 (0.7817)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [85]  [6000/6250]  eta: 0:01:07  lr: 0.000007  grad: 0.1971 (0.3642)  loss: 0.7852 (0.7818)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [85]  [6100/6250]  eta: 0:00:40  lr: 0.000007  grad: 0.1952 (0.3626)  loss: 0.7856 (0.7818)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [85]  [6200/6250]  eta: 0:00:13  lr: 0.000007  grad: 0.1942 (0.3615)  loss: 0.7831 (0.7818)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [85]  [6249/6250]  eta: 0:00:00  lr: 0.000007  grad: 0.1940 (0.3632)  loss: 0.7879 (0.7818)  time: 0.2698  data: 0.0002  max mem: 26157
Train: [85] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000007  grad: 0.1940 (0.3632)  loss: 0.7879 (0.7818)
Eval (hcp-train-subset): [85]  [ 0/62]  eta: 0:03:11  loss: 0.8007 (0.8007)  time: 3.0867  data: 2.9799  max mem: 26157
Eval (hcp-train-subset): [85]  [61/62]  eta: 0:00:00  loss: 0.7914 (0.7948)  time: 0.0995  data: 0.0151  max mem: 26157
Eval (hcp-train-subset): [85] Total time: 0:00:10 (0.1716 s / it)
Averaged stats (hcp-train-subset): loss: 0.7914 (0.7948)
Making plots (hcp-train-subset): example=46
Eval (hcp-val): [85]  [ 0/62]  eta: 0:03:21  loss: 0.8183 (0.8183)  time: 3.2504  data: 3.1478  max mem: 26157
Eval (hcp-val): [85]  [61/62]  eta: 0:00:00  loss: 0.8235 (0.8240)  time: 0.0878  data: 0.0051  max mem: 26157
Eval (hcp-val): [85] Total time: 0:00:11 (0.1782 s / it)
Averaged stats (hcp-val): loss: 0.8235 (0.8240)
Making plots (hcp-val): example=8
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [86]  [   0/6250]  eta: 9:04:10  lr: 0.000007  grad: 0.2156 (0.2156)  loss: 0.7899 (0.7899)  time: 5.2240  data: 4.9463  max mem: 26157
Train: [86]  [ 100/6250]  eta: 0:32:29  lr: 0.000007  grad: 0.2885 (0.2881)  loss: 0.7930 (0.8023)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [86]  [ 200/6250]  eta: 0:29:26  lr: 0.000007  grad: 0.2052 (0.3529)  loss: 0.7981 (0.7951)  time: 0.2658  data: 0.0001  max mem: 26157
Train: [86]  [ 300/6250]  eta: 0:28:07  lr: 0.000007  grad: 0.2289 (0.3668)  loss: 0.7812 (0.7921)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [86]  [ 400/6250]  eta: 0:27:15  lr: 0.000007  grad: 0.2289 (0.3875)  loss: 0.7826 (0.7900)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [86]  [ 500/6250]  eta: 0:26:33  lr: 0.000007  grad: 0.2404 (0.3759)  loss: 0.7819 (0.7887)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [86]  [ 600/6250]  eta: 0:25:56  lr: 0.000006  grad: 0.2165 (0.3819)  loss: 0.7891 (0.7879)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [ 700/6250]  eta: 0:25:21  lr: 0.000006  grad: 0.2075 (0.4107)  loss: 0.7833 (0.7879)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [86]  [ 800/6250]  eta: 0:24:49  lr: 0.000006  grad: 0.1980 (0.4160)  loss: 0.7860 (0.7880)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [86]  [ 900/6250]  eta: 0:24:18  lr: 0.000006  grad: 0.2285 (0.4092)  loss: 0.7813 (0.7879)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [86]  [1000/6250]  eta: 0:23:48  lr: 0.000006  grad: 0.2183 (0.3999)  loss: 0.7827 (0.7877)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [86]  [1100/6250]  eta: 0:23:18  lr: 0.000006  grad: 0.2110 (0.3830)  loss: 0.7855 (0.7874)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [86]  [1200/6250]  eta: 0:22:49  lr: 0.000006  grad: 0.2254 (0.3831)  loss: 0.7833 (0.7872)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [86]  [1300/6250]  eta: 0:22:20  lr: 0.000006  grad: 0.2014 (0.3727)  loss: 0.7844 (0.7869)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [1400/6250]  eta: 0:21:52  lr: 0.000006  grad: 0.2033 (0.3754)  loss: 0.7856 (0.7867)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [86]  [1500/6250]  eta: 0:21:24  lr: 0.000006  grad: 0.2142 (0.3682)  loss: 0.7825 (0.7867)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [86]  [1600/6250]  eta: 0:20:56  lr: 0.000006  grad: 0.2064 (0.3635)  loss: 0.7835 (0.7866)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [1700/6250]  eta: 0:20:28  lr: 0.000006  grad: 0.1970 (0.3558)  loss: 0.7877 (0.7865)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [86]  [1800/6250]  eta: 0:20:00  lr: 0.000006  grad: 0.1949 (0.3483)  loss: 0.7789 (0.7864)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [86]  [1900/6250]  eta: 0:19:32  lr: 0.000006  grad: 0.1981 (0.3473)  loss: 0.7817 (0.7863)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [86]  [2000/6250]  eta: 0:19:05  lr: 0.000006  grad: 0.1843 (0.3447)  loss: 0.7856 (0.7862)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [2100/6250]  eta: 0:18:37  lr: 0.000006  grad: 0.1910 (0.3442)  loss: 0.7842 (0.7861)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [86]  [2200/6250]  eta: 0:18:10  lr: 0.000006  grad: 0.1976 (0.3403)  loss: 0.7855 (0.7859)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [2300/6250]  eta: 0:17:42  lr: 0.000006  grad: 0.1951 (0.3362)  loss: 0.7814 (0.7858)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [2400/6250]  eta: 0:17:15  lr: 0.000006  grad: 0.2059 (0.3364)  loss: 0.7791 (0.7855)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [86]  [2500/6250]  eta: 0:16:48  lr: 0.000006  grad: 0.2082 (0.3366)  loss: 0.7807 (0.7854)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [2600/6250]  eta: 0:16:21  lr: 0.000006  grad: 0.2161 (0.3401)  loss: 0.7703 (0.7851)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [86]  [2700/6250]  eta: 0:15:54  lr: 0.000006  grad: 0.2117 (0.3380)  loss: 0.7784 (0.7849)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [86]  [2800/6250]  eta: 0:15:26  lr: 0.000006  grad: 0.2184 (0.3377)  loss: 0.7805 (0.7848)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [86]  [2900/6250]  eta: 0:14:59  lr: 0.000006  grad: 0.2256 (0.3414)  loss: 0.7752 (0.7846)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [86]  [3000/6250]  eta: 0:14:32  lr: 0.000006  grad: 0.2134 (0.3450)  loss: 0.7825 (0.7846)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [86]  [3100/6250]  eta: 0:14:05  lr: 0.000006  grad: 0.1999 (0.3444)  loss: 0.7851 (0.7845)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [86]  [3200/6250]  eta: 0:13:38  lr: 0.000006  grad: 0.2181 (0.3424)  loss: 0.7807 (0.7843)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [86]  [3300/6250]  eta: 0:13:11  lr: 0.000006  grad: 0.2177 (0.3454)  loss: 0.7827 (0.7844)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [3400/6250]  eta: 0:12:44  lr: 0.000006  grad: 0.2007 (0.3459)  loss: 0.7804 (0.7843)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [86]  [3500/6250]  eta: 0:12:17  lr: 0.000006  grad: 0.2208 (0.3445)  loss: 0.7763 (0.7843)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [3600/6250]  eta: 0:11:51  lr: 0.000006  grad: 0.2099 (0.3474)  loss: 0.7837 (0.7843)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [86]  [3700/6250]  eta: 0:11:24  lr: 0.000006  grad: 0.2291 (0.3470)  loss: 0.7822 (0.7843)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [86]  [3800/6250]  eta: 0:10:57  lr: 0.000006  grad: 0.2026 (0.3483)  loss: 0.7793 (0.7843)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [3900/6250]  eta: 0:10:30  lr: 0.000006  grad: 0.2048 (0.3481)  loss: 0.7811 (0.7842)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [4000/6250]  eta: 0:10:03  lr: 0.000006  grad: 0.2056 (0.3472)  loss: 0.7818 (0.7842)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [86]  [4100/6250]  eta: 0:09:36  lr: 0.000006  grad: 0.2175 (0.3453)  loss: 0.7775 (0.7842)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [86]  [4200/6250]  eta: 0:09:09  lr: 0.000006  grad: 0.2129 (0.3446)  loss: 0.7811 (0.7840)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [4300/6250]  eta: 0:08:42  lr: 0.000006  grad: 0.2193 (0.3535)  loss: 0.7778 (0.7840)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [86]  [4400/6250]  eta: 0:08:15  lr: 0.000006  grad: 0.2005 (0.3574)  loss: 0.7826 (0.7839)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [4500/6250]  eta: 0:07:49  lr: 0.000006  grad: 0.2092 (0.3587)  loss: 0.7815 (0.7840)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [86]  [4600/6250]  eta: 0:07:22  lr: 0.000006  grad: 0.1985 (0.3598)  loss: 0.7837 (0.7840)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [86]  [4700/6250]  eta: 0:06:55  lr: 0.000006  grad: 0.2132 (0.3589)  loss: 0.7835 (0.7840)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [86]  [4800/6250]  eta: 0:06:28  lr: 0.000006  grad: 0.1976 (0.3572)  loss: 0.7816 (0.7840)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [86]  [4900/6250]  eta: 0:06:01  lr: 0.000006  grad: 0.2039 (0.3553)  loss: 0.7830 (0.7840)  time: 0.2662  data: 0.0002  max mem: 26157
Train: [86]  [5000/6250]  eta: 0:05:34  lr: 0.000006  grad: 0.2010 (0.3549)  loss: 0.7841 (0.7840)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [86]  [5100/6250]  eta: 0:05:08  lr: 0.000006  grad: 0.1931 (0.3557)  loss: 0.7889 (0.7840)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [5200/6250]  eta: 0:04:41  lr: 0.000006  grad: 0.2014 (0.3541)  loss: 0.7833 (0.7840)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [86]  [5300/6250]  eta: 0:04:14  lr: 0.000006  grad: 0.2014 (0.3541)  loss: 0.7861 (0.7840)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [86]  [5400/6250]  eta: 0:03:47  lr: 0.000006  grad: 0.1914 (0.3544)  loss: 0.7752 (0.7840)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [86]  [5500/6250]  eta: 0:03:20  lr: 0.000006  grad: 0.2125 (0.3545)  loss: 0.7819 (0.7839)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [86]  [5600/6250]  eta: 0:02:54  lr: 0.000006  grad: 0.2168 (0.3546)  loss: 0.7839 (0.7839)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [86]  [5700/6250]  eta: 0:02:27  lr: 0.000006  grad: 0.2007 (0.3544)  loss: 0.7772 (0.7839)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [86]  [5800/6250]  eta: 0:02:00  lr: 0.000006  grad: 0.2026 (0.3542)  loss: 0.7832 (0.7839)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [86]  [5900/6250]  eta: 0:01:33  lr: 0.000006  grad: 0.2208 (0.3543)  loss: 0.7850 (0.7839)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [86]  [6000/6250]  eta: 0:01:06  lr: 0.000006  grad: 0.2165 (0.3543)  loss: 0.7773 (0.7838)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [86]  [6100/6250]  eta: 0:00:40  lr: 0.000006  grad: 0.2011 (0.3571)  loss: 0.7866 (0.7838)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [86]  [6200/6250]  eta: 0:00:13  lr: 0.000006  grad: 0.2085 (0.3596)  loss: 0.7794 (0.7838)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [86]  [6249/6250]  eta: 0:00:00  lr: 0.000006  grad: 0.2125 (0.3593)  loss: 0.7812 (0.7838)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [86] Total time: 0:27:56 (0.2682 s / it)
Averaged stats: lr: 0.000006  grad: 0.2125 (0.3593)  loss: 0.7812 (0.7838)
Eval (hcp-train-subset): [86]  [ 0/62]  eta: 0:03:21  loss: 0.8007 (0.8007)  time: 3.2424  data: 3.1408  max mem: 26157
Eval (hcp-train-subset): [86]  [61/62]  eta: 0:00:00  loss: 0.7906 (0.7926)  time: 0.0952  data: 0.0124  max mem: 26157
Eval (hcp-train-subset): [86] Total time: 0:00:11 (0.1795 s / it)
Averaged stats (hcp-train-subset): loss: 0.7906 (0.7926)
Making plots (hcp-train-subset): example=26
Eval (hcp-val): [86]  [ 0/62]  eta: 0:04:24  loss: 0.8207 (0.8207)  time: 4.2687  data: 4.1856  max mem: 26157
Eval (hcp-val): [86]  [61/62]  eta: 0:00:00  loss: 0.8233 (0.8241)  time: 0.0821  data: 0.0001  max mem: 26157
Eval (hcp-val): [86] Total time: 0:00:10 (0.1708 s / it)
Averaged stats (hcp-val): loss: 0.8233 (0.8241)
Making plots (hcp-val): example=18
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [87]  [   0/6250]  eta: 11:16:29  lr: 0.000006  grad: 0.1430 (0.1430)  loss: 0.8272 (0.8272)  time: 6.4943  data: 6.2235  max mem: 26157
Train: [87]  [ 100/6250]  eta: 0:33:47  lr: 0.000006  grad: 0.3013 (0.6362)  loss: 0.7544 (0.7798)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [87]  [ 200/6250]  eta: 0:30:07  lr: 0.000006  grad: 0.2107 (0.5314)  loss: 0.7786 (0.7785)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [87]  [ 300/6250]  eta: 0:28:34  lr: 0.000006  grad: 0.2003 (0.4664)  loss: 0.7939 (0.7819)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [87]  [ 400/6250]  eta: 0:27:34  lr: 0.000006  grad: 0.2146 (0.4459)  loss: 0.7851 (0.7831)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [87]  [ 500/6250]  eta: 0:26:48  lr: 0.000006  grad: 0.2339 (0.4349)  loss: 0.7768 (0.7822)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [87]  [ 600/6250]  eta: 0:26:08  lr: 0.000006  grad: 0.2142 (0.4241)  loss: 0.7832 (0.7820)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [87]  [ 700/6250]  eta: 0:25:32  lr: 0.000006  grad: 0.2139 (0.4278)  loss: 0.7761 (0.7819)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [ 800/6250]  eta: 0:24:58  lr: 0.000006  grad: 0.2040 (0.4405)  loss: 0.7909 (0.7820)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [87]  [ 900/6250]  eta: 0:24:26  lr: 0.000006  grad: 0.2044 (0.4371)  loss: 0.7818 (0.7824)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [87]  [1000/6250]  eta: 0:23:55  lr: 0.000006  grad: 0.1947 (0.4322)  loss: 0.7899 (0.7828)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [87]  [1100/6250]  eta: 0:23:25  lr: 0.000006  grad: 0.2184 (0.4149)  loss: 0.7765 (0.7830)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [87]  [1200/6250]  eta: 0:22:55  lr: 0.000006  grad: 0.2153 (0.4112)  loss: 0.7826 (0.7829)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [87]  [1300/6250]  eta: 0:22:26  lr: 0.000006  grad: 0.2210 (0.4074)  loss: 0.7839 (0.7830)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [87]  [1400/6250]  eta: 0:21:57  lr: 0.000005  grad: 0.1938 (0.4046)  loss: 0.7856 (0.7831)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [87]  [1500/6250]  eta: 0:21:28  lr: 0.000005  grad: 0.2122 (0.4030)  loss: 0.7803 (0.7830)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [87]  [1600/6250]  eta: 0:21:00  lr: 0.000005  grad: 0.1994 (0.4128)  loss: 0.7831 (0.7830)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [87]  [1700/6250]  eta: 0:20:32  lr: 0.000005  grad: 0.2110 (0.4034)  loss: 0.7807 (0.7831)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [87]  [1800/6250]  eta: 0:20:04  lr: 0.000005  grad: 0.2030 (0.4169)  loss: 0.7807 (0.7830)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [87]  [1900/6250]  eta: 0:19:36  lr: 0.000005  grad: 0.2059 (0.4114)  loss: 0.7800 (0.7830)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [2000/6250]  eta: 0:19:08  lr: 0.000005  grad: 0.2005 (0.4072)  loss: 0.7890 (0.7829)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [87]  [2100/6250]  eta: 0:18:40  lr: 0.000005  grad: 0.2072 (0.4109)  loss: 0.7845 (0.7829)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [2200/6250]  eta: 0:18:13  lr: 0.000005  grad: 0.1876 (0.4129)  loss: 0.7858 (0.7829)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [87]  [2300/6250]  eta: 0:17:45  lr: 0.000005  grad: 0.1984 (0.4140)  loss: 0.7787 (0.7829)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [87]  [2400/6250]  eta: 0:17:18  lr: 0.000005  grad: 0.2349 (0.4092)  loss: 0.7865 (0.7831)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [2500/6250]  eta: 0:16:50  lr: 0.000005  grad: 0.1978 (0.4079)  loss: 0.7800 (0.7831)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [87]  [2600/6250]  eta: 0:16:23  lr: 0.000005  grad: 0.1999 (0.4194)  loss: 0.7836 (0.7831)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [87]  [2700/6250]  eta: 0:15:56  lr: 0.000005  grad: 0.2018 (0.4162)  loss: 0.7822 (0.7831)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [87]  [2800/6250]  eta: 0:15:29  lr: 0.000005  grad: 0.1847 (0.4156)  loss: 0.7895 (0.7832)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [87]  [2900/6250]  eta: 0:15:01  lr: 0.000005  grad: 0.2052 (0.4142)  loss: 0.7751 (0.7832)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [87]  [3000/6250]  eta: 0:14:34  lr: 0.000005  grad: 0.2031 (0.4156)  loss: 0.7799 (0.7832)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [87]  [3100/6250]  eta: 0:14:07  lr: 0.000005  grad: 0.2031 (0.4143)  loss: 0.7768 (0.7831)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [87]  [3200/6250]  eta: 0:13:40  lr: 0.000005  grad: 0.2040 (0.4185)  loss: 0.7835 (0.7831)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [87]  [3300/6250]  eta: 0:13:13  lr: 0.000005  grad: 0.1993 (0.4196)  loss: 0.7824 (0.7831)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [3400/6250]  eta: 0:12:46  lr: 0.000005  grad: 0.2123 (0.4275)  loss: 0.7890 (0.7832)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [87]  [3500/6250]  eta: 0:12:19  lr: 0.000005  grad: 0.1994 (0.4267)  loss: 0.7885 (0.7832)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [87]  [3600/6250]  eta: 0:11:52  lr: 0.000005  grad: 0.2003 (0.4260)  loss: 0.7847 (0.7833)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [87]  [3700/6250]  eta: 0:11:25  lr: 0.000005  grad: 0.1942 (0.4257)  loss: 0.7905 (0.7833)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [87]  [3800/6250]  eta: 0:10:58  lr: 0.000005  grad: 0.2107 (0.4300)  loss: 0.7777 (0.7833)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [87]  [3900/6250]  eta: 0:10:31  lr: 0.000005  grad: 0.2118 (0.4321)  loss: 0.7707 (0.7833)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [87]  [4000/6250]  eta: 0:10:04  lr: 0.000005  grad: 0.1895 (0.4294)  loss: 0.7801 (0.7831)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [87]  [4100/6250]  eta: 0:09:37  lr: 0.000005  grad: 0.2062 (0.4317)  loss: 0.7774 (0.7831)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [87]  [4200/6250]  eta: 0:09:10  lr: 0.000005  grad: 0.1986 (0.4297)  loss: 0.7775 (0.7830)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [87]  [4300/6250]  eta: 0:08:43  lr: 0.000005  grad: 0.2099 (0.4277)  loss: 0.7781 (0.7829)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [87]  [4400/6250]  eta: 0:08:16  lr: 0.000005  grad: 0.2060 (0.4251)  loss: 0.7783 (0.7828)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [87]  [4500/6250]  eta: 0:07:49  lr: 0.000005  grad: 0.2076 (0.4235)  loss: 0.7778 (0.7827)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [87]  [4600/6250]  eta: 0:07:22  lr: 0.000005  grad: 0.2019 (0.4256)  loss: 0.7782 (0.7826)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [87]  [4700/6250]  eta: 0:06:55  lr: 0.000005  grad: 0.2238 (0.4256)  loss: 0.7750 (0.7824)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [4800/6250]  eta: 0:06:28  lr: 0.000005  grad: 0.2085 (0.4236)  loss: 0.7758 (0.7824)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [87]  [4900/6250]  eta: 0:06:02  lr: 0.000005  grad: 0.2116 (0.4276)  loss: 0.7779 (0.7822)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [87]  [5000/6250]  eta: 0:05:35  lr: 0.000005  grad: 0.2240 (0.4328)  loss: 0.7732 (0.7821)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [87]  [5100/6250]  eta: 0:05:08  lr: 0.000005  grad: 0.2105 (0.4332)  loss: 0.7754 (0.7819)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [87]  [5200/6250]  eta: 0:04:41  lr: 0.000005  grad: 0.2196 (0.4375)  loss: 0.7808 (0.7819)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [5300/6250]  eta: 0:04:14  lr: 0.000005  grad: 0.2105 (0.4350)  loss: 0.7783 (0.7818)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [87]  [5400/6250]  eta: 0:03:47  lr: 0.000005  grad: 0.1968 (0.4358)  loss: 0.7877 (0.7818)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [87]  [5500/6250]  eta: 0:03:21  lr: 0.000005  grad: 0.2229 (0.4394)  loss: 0.7814 (0.7818)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [87]  [5600/6250]  eta: 0:02:54  lr: 0.000005  grad: 0.2210 (0.4421)  loss: 0.7804 (0.7818)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [87]  [5700/6250]  eta: 0:02:27  lr: 0.000005  grad: 0.1988 (0.4415)  loss: 0.7788 (0.7818)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [87]  [5800/6250]  eta: 0:02:00  lr: 0.000005  grad: 0.2215 (0.4386)  loss: 0.7810 (0.7817)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [87]  [5900/6250]  eta: 0:01:33  lr: 0.000005  grad: 0.2182 (0.4370)  loss: 0.7787 (0.7816)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [87]  [6000/6250]  eta: 0:01:07  lr: 0.000005  grad: 0.2133 (0.4369)  loss: 0.7750 (0.7815)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [87]  [6100/6250]  eta: 0:00:40  lr: 0.000005  grad: 0.2045 (0.4373)  loss: 0.7723 (0.7814)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [87]  [6200/6250]  eta: 0:00:13  lr: 0.000005  grad: 0.2120 (0.4354)  loss: 0.7718 (0.7813)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [87]  [6249/6250]  eta: 0:00:00  lr: 0.000005  grad: 0.2090 (0.4339)  loss: 0.7788 (0.7813)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [87] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000005  grad: 0.2090 (0.4339)  loss: 0.7788 (0.7813)
Eval (hcp-train-subset): [87]  [ 0/62]  eta: 0:04:07  loss: 0.8006 (0.8006)  time: 3.9982  data: 3.9156  max mem: 26157
Eval (hcp-train-subset): [87]  [61/62]  eta: 0:00:00  loss: 0.7900 (0.7913)  time: 0.0853  data: 0.0031  max mem: 26157
Eval (hcp-train-subset): [87] Total time: 0:00:10 (0.1643 s / it)
Averaged stats (hcp-train-subset): loss: 0.7900 (0.7913)
Making plots (hcp-train-subset): example=22
Eval (hcp-val): [87]  [ 0/62]  eta: 0:04:34  loss: 0.8216 (0.8216)  time: 4.4227  data: 4.3397  max mem: 26157
Eval (hcp-val): [87]  [61/62]  eta: 0:00:00  loss: 0.8239 (0.8240)  time: 0.0852  data: 0.0027  max mem: 26157
Eval (hcp-val): [87] Total time: 0:00:10 (0.1711 s / it)
Averaged stats (hcp-val): loss: 0.8239 (0.8240)
Making plots (hcp-val): example=28
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [88]  [   0/6250]  eta: 8:57:43  lr: 0.000005  grad: 0.2054 (0.2054)  loss: 0.8131 (0.8131)  time: 5.1622  data: 4.8933  max mem: 26157
Train: [88]  [ 100/6250]  eta: 0:32:27  lr: 0.000005  grad: 0.2469 (0.3119)  loss: 0.7929 (0.7917)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [88]  [ 200/6250]  eta: 0:29:27  lr: 0.000005  grad: 0.2370 (0.4498)  loss: 0.7698 (0.7840)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [88]  [ 300/6250]  eta: 0:28:09  lr: 0.000005  grad: 0.2190 (0.4061)  loss: 0.7789 (0.7828)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [88]  [ 400/6250]  eta: 0:27:16  lr: 0.000005  grad: 0.2137 (0.4354)  loss: 0.7768 (0.7826)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [88]  [ 500/6250]  eta: 0:26:34  lr: 0.000005  grad: 0.1988 (0.4086)  loss: 0.7898 (0.7830)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [88]  [ 600/6250]  eta: 0:25:56  lr: 0.000005  grad: 0.2108 (0.3953)  loss: 0.7816 (0.7833)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [88]  [ 700/6250]  eta: 0:25:22  lr: 0.000005  grad: 0.2102 (0.3765)  loss: 0.7891 (0.7837)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [88]  [ 800/6250]  eta: 0:24:50  lr: 0.000005  grad: 0.2130 (0.3679)  loss: 0.7891 (0.7840)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [88]  [ 900/6250]  eta: 0:24:19  lr: 0.000005  grad: 0.2024 (0.3609)  loss: 0.7845 (0.7840)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [88]  [1000/6250]  eta: 0:23:49  lr: 0.000005  grad: 0.2127 (0.3645)  loss: 0.7843 (0.7843)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [88]  [1100/6250]  eta: 0:23:19  lr: 0.000005  grad: 0.1998 (0.3711)  loss: 0.7877 (0.7842)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [88]  [1200/6250]  eta: 0:22:50  lr: 0.000005  grad: 0.2011 (0.3757)  loss: 0.7869 (0.7842)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [88]  [1300/6250]  eta: 0:22:21  lr: 0.000005  grad: 0.1936 (0.3770)  loss: 0.7872 (0.7841)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [88]  [1400/6250]  eta: 0:21:53  lr: 0.000005  grad: 0.2050 (0.4220)  loss: 0.7761 (0.7839)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [88]  [1500/6250]  eta: 0:21:24  lr: 0.000005  grad: 0.2026 (0.4247)  loss: 0.7838 (0.7835)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [88]  [1600/6250]  eta: 0:20:56  lr: 0.000005  grad: 0.2004 (0.4200)  loss: 0.7793 (0.7834)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [88]  [1700/6250]  eta: 0:20:28  lr: 0.000005  grad: 0.2157 (0.4251)  loss: 0.7795 (0.7830)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [88]  [1800/6250]  eta: 0:20:01  lr: 0.000005  grad: 0.2080 (0.4265)  loss: 0.7770 (0.7828)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [88]  [1900/6250]  eta: 0:19:33  lr: 0.000005  grad: 0.2062 (0.4284)  loss: 0.7823 (0.7827)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [88]  [2000/6250]  eta: 0:19:05  lr: 0.000005  grad: 0.1998 (0.4349)  loss: 0.7784 (0.7826)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [88]  [2100/6250]  eta: 0:18:38  lr: 0.000005  grad: 0.2039 (0.4275)  loss: 0.7823 (0.7827)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [88]  [2200/6250]  eta: 0:18:11  lr: 0.000005  grad: 0.1980 (0.4217)  loss: 0.7911 (0.7830)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [88]  [2300/6250]  eta: 0:17:43  lr: 0.000005  grad: 0.2151 (0.4152)  loss: 0.7767 (0.7830)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [88]  [2400/6250]  eta: 0:17:16  lr: 0.000005  grad: 0.2039 (0.4098)  loss: 0.7933 (0.7832)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [88]  [2500/6250]  eta: 0:16:49  lr: 0.000005  grad: 0.1939 (0.4078)  loss: 0.7890 (0.7833)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [88]  [2600/6250]  eta: 0:16:21  lr: 0.000005  grad: 0.2028 (0.4031)  loss: 0.7863 (0.7835)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [88]  [2700/6250]  eta: 0:15:54  lr: 0.000005  grad: 0.2150 (0.3984)  loss: 0.7845 (0.7836)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [88]  [2800/6250]  eta: 0:15:27  lr: 0.000005  grad: 0.2063 (0.3936)  loss: 0.7861 (0.7836)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [88]  [2900/6250]  eta: 0:15:00  lr: 0.000004  grad: 0.2076 (0.3966)  loss: 0.7874 (0.7837)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [88]  [3000/6250]  eta: 0:14:33  lr: 0.000004  grad: 0.2177 (0.3921)  loss: 0.7763 (0.7838)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [88]  [3100/6250]  eta: 0:14:06  lr: 0.000004  grad: 0.2093 (0.3903)  loss: 0.7861 (0.7839)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [88]  [3200/6250]  eta: 0:13:39  lr: 0.000004  grad: 0.2196 (0.3902)  loss: 0.7815 (0.7839)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [88]  [3300/6250]  eta: 0:13:12  lr: 0.000004  grad: 0.2062 (0.3917)  loss: 0.7788 (0.7840)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [88]  [3400/6250]  eta: 0:12:45  lr: 0.000004  grad: 0.2009 (0.3901)  loss: 0.7878 (0.7840)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [88]  [3500/6250]  eta: 0:12:18  lr: 0.000004  grad: 0.2152 (0.3895)  loss: 0.7831 (0.7839)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [88]  [3600/6250]  eta: 0:11:51  lr: 0.000004  grad: 0.2165 (0.3917)  loss: 0.7812 (0.7839)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [88]  [3700/6250]  eta: 0:11:24  lr: 0.000004  grad: 0.2211 (0.3894)  loss: 0.7842 (0.7838)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [88]  [3800/6250]  eta: 0:10:57  lr: 0.000004  grad: 0.2010 (0.3890)  loss: 0.7750 (0.7838)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [88]  [3900/6250]  eta: 0:10:30  lr: 0.000004  grad: 0.2098 (0.3873)  loss: 0.7783 (0.7837)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [88]  [4000/6250]  eta: 0:10:03  lr: 0.000004  grad: 0.2024 (0.3862)  loss: 0.7804 (0.7836)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [88]  [4100/6250]  eta: 0:09:36  lr: 0.000004  grad: 0.2036 (0.3865)  loss: 0.7773 (0.7835)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [88]  [4200/6250]  eta: 0:09:09  lr: 0.000004  grad: 0.2092 (0.3857)  loss: 0.7836 (0.7834)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [88]  [4300/6250]  eta: 0:08:42  lr: 0.000004  grad: 0.2148 (0.3869)  loss: 0.7761 (0.7833)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [88]  [4400/6250]  eta: 0:08:16  lr: 0.000004  grad: 0.2019 (0.3902)  loss: 0.7822 (0.7832)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [88]  [4500/6250]  eta: 0:07:49  lr: 0.000004  grad: 0.2156 (0.3927)  loss: 0.7707 (0.7831)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [88]  [4600/6250]  eta: 0:07:22  lr: 0.000004  grad: 0.2126 (0.3953)  loss: 0.7745 (0.7830)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [88]  [4700/6250]  eta: 0:06:55  lr: 0.000004  grad: 0.2039 (0.3934)  loss: 0.7827 (0.7829)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [88]  [4800/6250]  eta: 0:06:28  lr: 0.000004  grad: 0.2096 (0.3903)  loss: 0.7857 (0.7829)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [88]  [4900/6250]  eta: 0:06:01  lr: 0.000004  grad: 0.2061 (0.3882)  loss: 0.7916 (0.7828)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [88]  [5000/6250]  eta: 0:05:34  lr: 0.000004  grad: 0.2112 (0.3865)  loss: 0.7850 (0.7828)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [88]  [5100/6250]  eta: 0:05:08  lr: 0.000004  grad: 0.2102 (0.3843)  loss: 0.7862 (0.7828)  time: 0.2660  data: 0.0002  max mem: 26157
Train: [88]  [5200/6250]  eta: 0:04:41  lr: 0.000004  grad: 0.2140 (0.3865)  loss: 0.7769 (0.7828)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [88]  [5300/6250]  eta: 0:04:14  lr: 0.000004  grad: 0.2083 (0.3898)  loss: 0.7759 (0.7827)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [88]  [5400/6250]  eta: 0:03:47  lr: 0.000004  grad: 0.2342 (0.3902)  loss: 0.7795 (0.7827)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [88]  [5500/6250]  eta: 0:03:20  lr: 0.000004  grad: 0.1999 (0.3890)  loss: 0.7817 (0.7826)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [88]  [5600/6250]  eta: 0:02:54  lr: 0.000004  grad: 0.2133 (0.3865)  loss: 0.7788 (0.7826)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [88]  [5700/6250]  eta: 0:02:27  lr: 0.000004  grad: 0.2049 (0.3858)  loss: 0.7772 (0.7826)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [88]  [5800/6250]  eta: 0:02:00  lr: 0.000004  grad: 0.2036 (0.3872)  loss: 0.7801 (0.7826)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [88]  [5900/6250]  eta: 0:01:33  lr: 0.000004  grad: 0.1936 (0.3863)  loss: 0.7879 (0.7826)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [88]  [6000/6250]  eta: 0:01:06  lr: 0.000004  grad: 0.2162 (0.3846)  loss: 0.7737 (0.7826)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [88]  [6100/6250]  eta: 0:00:40  lr: 0.000004  grad: 0.2133 (0.3839)  loss: 0.7831 (0.7826)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [88]  [6200/6250]  eta: 0:00:13  lr: 0.000004  grad: 0.1954 (0.3872)  loss: 0.7881 (0.7826)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [88]  [6249/6250]  eta: 0:00:00  lr: 0.000004  grad: 0.2007 (0.3865)  loss: 0.7816 (0.7826)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [88] Total time: 0:27:57 (0.2683 s / it)
Averaged stats: lr: 0.000004  grad: 0.2007 (0.3865)  loss: 0.7816 (0.7826)
Eval (hcp-train-subset): [88]  [ 0/62]  eta: 0:03:36  loss: 0.7942 (0.7942)  time: 3.4842  data: 3.3710  max mem: 26157
Eval (hcp-train-subset): [88]  [61/62]  eta: 0:00:00  loss: 0.7857 (0.7888)  time: 0.0823  data: 0.0001  max mem: 26157
Eval (hcp-train-subset): [88] Total time: 0:00:10 (0.1689 s / it)
Averaged stats (hcp-train-subset): loss: 0.7857 (0.7888)
Making plots (hcp-train-subset): example=10
Eval (hcp-val): [88]  [ 0/62]  eta: 0:04:47  loss: 0.8206 (0.8206)  time: 4.6381  data: 4.5553  max mem: 26157
Eval (hcp-val): [88]  [61/62]  eta: 0:00:00  loss: 0.8222 (0.8230)  time: 0.1171  data: 0.0345  max mem: 26157
Eval (hcp-val): [88] Total time: 0:00:11 (0.1820 s / it)
Averaged stats (hcp-val): loss: 0.8222 (0.8230)
Making plots (hcp-val): example=29
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [89]  [   0/6250]  eta: 8:23:11  lr: 0.000004  grad: 0.2737 (0.2737)  loss: 0.8347 (0.8347)  time: 4.8307  data: 4.5478  max mem: 26157
Train: [89]  [ 100/6250]  eta: 0:32:46  lr: 0.000004  grad: 0.2297 (0.3430)  loss: 0.8056 (0.7989)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [89]  [ 200/6250]  eta: 0:29:38  lr: 0.000004  grad: 0.2435 (0.3447)  loss: 0.7752 (0.7900)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [89]  [ 300/6250]  eta: 0:28:17  lr: 0.000004  grad: 0.2206 (0.3235)  loss: 0.7934 (0.7887)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [89]  [ 400/6250]  eta: 0:27:21  lr: 0.000004  grad: 0.2509 (0.3409)  loss: 0.7793 (0.7869)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [89]  [ 500/6250]  eta: 0:26:37  lr: 0.000004  grad: 0.2261 (0.3403)  loss: 0.7755 (0.7859)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [89]  [ 600/6250]  eta: 0:25:59  lr: 0.000004  grad: 0.2411 (0.3443)  loss: 0.7823 (0.7856)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [89]  [ 700/6250]  eta: 0:25:24  lr: 0.000004  grad: 0.2092 (0.3705)  loss: 0.7795 (0.7847)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [89]  [ 800/6250]  eta: 0:24:52  lr: 0.000004  grad: 0.2312 (0.3567)  loss: 0.7775 (0.7837)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [89]  [ 900/6250]  eta: 0:24:21  lr: 0.000004  grad: 0.2224 (0.3469)  loss: 0.7716 (0.7831)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [89]  [1000/6250]  eta: 0:23:50  lr: 0.000004  grad: 0.2129 (0.3546)  loss: 0.7815 (0.7827)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [89]  [1100/6250]  eta: 0:23:20  lr: 0.000004  grad: 0.2029 (0.3556)  loss: 0.7825 (0.7825)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [89]  [1200/6250]  eta: 0:22:51  lr: 0.000004  grad: 0.2047 (0.3750)  loss: 0.7856 (0.7824)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [89]  [1300/6250]  eta: 0:22:22  lr: 0.000004  grad: 0.2101 (0.3679)  loss: 0.7814 (0.7821)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [89]  [1400/6250]  eta: 0:21:54  lr: 0.000004  grad: 0.2078 (0.3651)  loss: 0.7821 (0.7819)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [89]  [1500/6250]  eta: 0:21:25  lr: 0.000004  grad: 0.2079 (0.3731)  loss: 0.7787 (0.7816)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [89]  [1600/6250]  eta: 0:20:57  lr: 0.000004  grad: 0.2145 (0.3660)  loss: 0.7815 (0.7816)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [89]  [1700/6250]  eta: 0:20:29  lr: 0.000004  grad: 0.2037 (0.3633)  loss: 0.7789 (0.7814)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [89]  [1800/6250]  eta: 0:20:01  lr: 0.000004  grad: 0.2256 (0.3631)  loss: 0.7750 (0.7812)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [89]  [1900/6250]  eta: 0:19:34  lr: 0.000004  grad: 0.1991 (0.3793)  loss: 0.7856 (0.7813)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [89]  [2000/6250]  eta: 0:19:06  lr: 0.000004  grad: 0.2087 (0.3730)  loss: 0.7804 (0.7811)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [89]  [2100/6250]  eta: 0:18:39  lr: 0.000004  grad: 0.2081 (0.3835)  loss: 0.7758 (0.7809)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [89]  [2200/6250]  eta: 0:18:11  lr: 0.000004  grad: 0.2154 (0.3772)  loss: 0.7828 (0.7808)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [89]  [2300/6250]  eta: 0:17:44  lr: 0.000004  grad: 0.2014 (0.3736)  loss: 0.7792 (0.7807)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [89]  [2400/6250]  eta: 0:17:16  lr: 0.000004  grad: 0.2141 (0.3708)  loss: 0.7822 (0.7807)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [89]  [2500/6250]  eta: 0:16:49  lr: 0.000004  grad: 0.2033 (0.3683)  loss: 0.7772 (0.7806)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [89]  [2600/6250]  eta: 0:16:22  lr: 0.000004  grad: 0.2218 (0.3648)  loss: 0.7750 (0.7805)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [89]  [2700/6250]  eta: 0:15:55  lr: 0.000004  grad: 0.2143 (0.3636)  loss: 0.7781 (0.7804)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [89]  [2800/6250]  eta: 0:15:28  lr: 0.000004  grad: 0.2066 (0.3599)  loss: 0.7800 (0.7803)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [89]  [2900/6250]  eta: 0:15:00  lr: 0.000004  grad: 0.2050 (0.3605)  loss: 0.7860 (0.7803)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [89]  [3000/6250]  eta: 0:14:33  lr: 0.000004  grad: 0.2116 (0.3716)  loss: 0.7803 (0.7803)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [89]  [3100/6250]  eta: 0:14:06  lr: 0.000004  grad: 0.2052 (0.3686)  loss: 0.7773 (0.7802)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [89]  [3200/6250]  eta: 0:13:39  lr: 0.000004  grad: 0.2125 (0.3673)  loss: 0.7863 (0.7803)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [89]  [3300/6250]  eta: 0:13:12  lr: 0.000004  grad: 0.2135 (0.3657)  loss: 0.7725 (0.7802)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [89]  [3400/6250]  eta: 0:12:45  lr: 0.000004  grad: 0.2059 (0.3726)  loss: 0.7782 (0.7801)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [89]  [3500/6250]  eta: 0:12:18  lr: 0.000004  grad: 0.2067 (0.3760)  loss: 0.7804 (0.7801)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [89]  [3600/6250]  eta: 0:11:51  lr: 0.000004  grad: 0.2005 (0.3725)  loss: 0.7842 (0.7801)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [89]  [3700/6250]  eta: 0:11:24  lr: 0.000004  grad: 0.2239 (0.3695)  loss: 0.7762 (0.7802)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [89]  [3800/6250]  eta: 0:10:57  lr: 0.000004  grad: 0.2148 (0.3682)  loss: 0.7783 (0.7801)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [89]  [3900/6250]  eta: 0:10:30  lr: 0.000004  grad: 0.1975 (0.3675)  loss: 0.7825 (0.7802)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [89]  [4000/6250]  eta: 0:10:03  lr: 0.000004  grad: 0.2237 (0.3669)  loss: 0.7748 (0.7801)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [89]  [4100/6250]  eta: 0:09:37  lr: 0.000004  grad: 0.2143 (0.3682)  loss: 0.7795 (0.7801)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [89]  [4200/6250]  eta: 0:09:10  lr: 0.000004  grad: 0.2061 (0.3665)  loss: 0.7830 (0.7801)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [89]  [4300/6250]  eta: 0:08:43  lr: 0.000004  grad: 0.2248 (0.3659)  loss: 0.7808 (0.7801)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [89]  [4400/6250]  eta: 0:08:16  lr: 0.000004  grad: 0.2271 (0.3708)  loss: 0.7761 (0.7802)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [89]  [4500/6250]  eta: 0:07:49  lr: 0.000004  grad: 0.2101 (0.3727)  loss: 0.7842 (0.7803)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [89]  [4600/6250]  eta: 0:07:22  lr: 0.000004  grad: 0.2057 (0.3727)  loss: 0.7882 (0.7803)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [89]  [4700/6250]  eta: 0:06:55  lr: 0.000004  grad: 0.2180 (0.3752)  loss: 0.7817 (0.7804)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [89]  [4800/6250]  eta: 0:06:28  lr: 0.000004  grad: 0.2090 (0.3761)  loss: 0.7835 (0.7805)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [89]  [4900/6250]  eta: 0:06:02  lr: 0.000004  grad: 0.2141 (0.3749)  loss: 0.7790 (0.7806)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [89]  [5000/6250]  eta: 0:05:35  lr: 0.000004  grad: 0.1925 (0.3741)  loss: 0.7895 (0.7807)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [89]  [5100/6250]  eta: 0:05:08  lr: 0.000004  grad: 0.2124 (0.3751)  loss: 0.7733 (0.7807)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [89]  [5200/6250]  eta: 0:04:41  lr: 0.000003  grad: 0.2136 (0.3759)  loss: 0.7848 (0.7808)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [89]  [5300/6250]  eta: 0:04:14  lr: 0.000003  grad: 0.2231 (0.3761)  loss: 0.7805 (0.7809)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [89]  [5400/6250]  eta: 0:03:47  lr: 0.000003  grad: 0.2047 (0.3764)  loss: 0.7835 (0.7809)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [89]  [5500/6250]  eta: 0:03:21  lr: 0.000003  grad: 0.2093 (0.3751)  loss: 0.7810 (0.7809)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [89]  [5600/6250]  eta: 0:02:54  lr: 0.000003  grad: 0.2127 (0.3742)  loss: 0.7842 (0.7810)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [89]  [5700/6250]  eta: 0:02:27  lr: 0.000003  grad: 0.2097 (0.3735)  loss: 0.7827 (0.7811)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [89]  [5800/6250]  eta: 0:02:00  lr: 0.000003  grad: 0.2165 (0.3740)  loss: 0.7827 (0.7812)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [89]  [5900/6250]  eta: 0:01:33  lr: 0.000003  grad: 0.1997 (0.3747)  loss: 0.7858 (0.7813)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [89]  [6000/6250]  eta: 0:01:07  lr: 0.000003  grad: 0.2113 (0.3734)  loss: 0.7904 (0.7813)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [89]  [6100/6250]  eta: 0:00:40  lr: 0.000003  grad: 0.2167 (0.3745)  loss: 0.7821 (0.7814)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [89]  [6200/6250]  eta: 0:00:13  lr: 0.000003  grad: 0.2065 (0.3739)  loss: 0.7856 (0.7814)  time: 0.2662  data: 0.0002  max mem: 26157
Train: [89]  [6249/6250]  eta: 0:00:00  lr: 0.000003  grad: 0.2013 (0.3733)  loss: 0.7838 (0.7814)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [89] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000003  grad: 0.2013 (0.3733)  loss: 0.7838 (0.7814)
Eval (hcp-train-subset): [89]  [ 0/62]  eta: 0:04:29  loss: 0.7939 (0.7939)  time: 4.3428  data: 4.2602  max mem: 26157
Eval (hcp-train-subset): [89]  [61/62]  eta: 0:00:00  loss: 0.7841 (0.7885)  time: 0.0893  data: 0.0052  max mem: 26157
Eval (hcp-train-subset): [89] Total time: 0:00:10 (0.1678 s / it)
Averaged stats (hcp-train-subset): loss: 0.7841 (0.7885)
Making plots (hcp-train-subset): example=54
Eval (hcp-val): [89]  [ 0/62]  eta: 0:02:55  loss: 0.8190 (0.8190)  time: 2.8236  data: 2.7240  max mem: 26157
Eval (hcp-val): [89]  [61/62]  eta: 0:00:00  loss: 0.8217 (0.8232)  time: 0.0878  data: 0.0054  max mem: 26157
Eval (hcp-val): [89] Total time: 0:00:10 (0.1705 s / it)
Averaged stats (hcp-val): loss: 0.8217 (0.8232)
Making plots (hcp-val): example=33
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [90]  [   0/6250]  eta: 8:46:12  lr: 0.000003  grad: 0.3723 (0.3723)  loss: 0.7716 (0.7716)  time: 5.0517  data: 4.7804  max mem: 26157
Train: [90]  [ 100/6250]  eta: 0:32:21  lr: 0.000003  grad: 0.2862 (0.4008)  loss: 0.7929 (0.7899)  time: 0.2688  data: 0.0001  max mem: 26157
Train: [90]  [ 200/6250]  eta: 0:29:27  lr: 0.000003  grad: 0.2464 (0.3802)  loss: 0.7755 (0.7873)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [90]  [ 300/6250]  eta: 0:28:09  lr: 0.000003  grad: 0.2263 (0.3897)  loss: 0.7899 (0.7860)  time: 0.2708  data: 0.0002  max mem: 26157
Train: [90]  [ 400/6250]  eta: 0:27:16  lr: 0.000003  grad: 0.2346 (0.3686)  loss: 0.7918 (0.7869)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [90]  [ 500/6250]  eta: 0:26:34  lr: 0.000003  grad: 0.2228 (0.3657)  loss: 0.7893 (0.7871)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [90]  [ 600/6250]  eta: 0:25:57  lr: 0.000003  grad: 0.2164 (0.3497)  loss: 0.7901 (0.7869)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [90]  [ 700/6250]  eta: 0:25:23  lr: 0.000003  grad: 0.2209 (0.3413)  loss: 0.7776 (0.7868)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [ 800/6250]  eta: 0:24:51  lr: 0.000003  grad: 0.2125 (0.3403)  loss: 0.7808 (0.7863)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [90]  [ 900/6250]  eta: 0:24:20  lr: 0.000003  grad: 0.2065 (0.3532)  loss: 0.7782 (0.7863)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [90]  [1000/6250]  eta: 0:23:49  lr: 0.000003  grad: 0.2149 (0.3572)  loss: 0.7820 (0.7858)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [1100/6250]  eta: 0:23:20  lr: 0.000003  grad: 0.1949 (0.3696)  loss: 0.7891 (0.7855)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [90]  [1200/6250]  eta: 0:22:51  lr: 0.000003  grad: 0.2105 (0.3632)  loss: 0.7900 (0.7853)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [90]  [1300/6250]  eta: 0:22:22  lr: 0.000003  grad: 0.1931 (0.3715)  loss: 0.7891 (0.7854)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [1400/6250]  eta: 0:21:53  lr: 0.000003  grad: 0.2245 (0.3665)  loss: 0.7837 (0.7853)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [90]  [1500/6250]  eta: 0:21:25  lr: 0.000003  grad: 0.2048 (0.3642)  loss: 0.7853 (0.7853)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [1600/6250]  eta: 0:20:57  lr: 0.000003  grad: 0.2159 (0.3616)  loss: 0.7819 (0.7851)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [90]  [1700/6250]  eta: 0:20:29  lr: 0.000003  grad: 0.2118 (0.3644)  loss: 0.7840 (0.7849)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [90]  [1800/6250]  eta: 0:20:01  lr: 0.000003  grad: 0.2137 (0.3625)  loss: 0.7775 (0.7847)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [90]  [1900/6250]  eta: 0:19:33  lr: 0.000003  grad: 0.2140 (0.3607)  loss: 0.7771 (0.7844)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [90]  [2000/6250]  eta: 0:19:06  lr: 0.000003  grad: 0.2141 (0.3559)  loss: 0.7792 (0.7844)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [90]  [2100/6250]  eta: 0:18:38  lr: 0.000003  grad: 0.2075 (0.3601)  loss: 0.7835 (0.7842)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [2200/6250]  eta: 0:18:11  lr: 0.000003  grad: 0.2100 (0.3569)  loss: 0.7878 (0.7843)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [2300/6250]  eta: 0:17:43  lr: 0.000003  grad: 0.2112 (0.3557)  loss: 0.7817 (0.7842)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [2400/6250]  eta: 0:17:16  lr: 0.000003  grad: 0.2209 (0.3607)  loss: 0.7827 (0.7842)  time: 0.2658  data: 0.0001  max mem: 26157
Train: [90]  [2500/6250]  eta: 0:16:49  lr: 0.000003  grad: 0.2050 (0.3619)  loss: 0.7809 (0.7841)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [2600/6250]  eta: 0:16:22  lr: 0.000003  grad: 0.2190 (0.3613)  loss: 0.7851 (0.7841)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [90]  [2700/6250]  eta: 0:15:54  lr: 0.000003  grad: 0.2107 (0.3658)  loss: 0.7803 (0.7842)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [90]  [2800/6250]  eta: 0:15:27  lr: 0.000003  grad: 0.2118 (0.3634)  loss: 0.7812 (0.7842)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [2900/6250]  eta: 0:15:00  lr: 0.000003  grad: 0.2041 (0.3589)  loss: 0.7838 (0.7842)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [3000/6250]  eta: 0:14:33  lr: 0.000003  grad: 0.2305 (0.3659)  loss: 0.7873 (0.7842)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [3100/6250]  eta: 0:14:06  lr: 0.000003  grad: 0.2119 (0.3640)  loss: 0.7803 (0.7841)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [90]  [3200/6250]  eta: 0:13:39  lr: 0.000003  grad: 0.2147 (0.3673)  loss: 0.7817 (0.7841)  time: 0.2660  data: 0.0002  max mem: 26157
Train: [90]  [3300/6250]  eta: 0:13:12  lr: 0.000003  grad: 0.2130 (0.3659)  loss: 0.7856 (0.7841)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [90]  [3400/6250]  eta: 0:12:45  lr: 0.000003  grad: 0.2090 (0.3654)  loss: 0.7857 (0.7840)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [90]  [3500/6250]  eta: 0:12:18  lr: 0.000003  grad: 0.2281 (0.3676)  loss: 0.7835 (0.7840)  time: 0.2707  data: 0.0002  max mem: 26157
Train: [90]  [3600/6250]  eta: 0:11:51  lr: 0.000003  grad: 0.2149 (0.3686)  loss: 0.7839 (0.7839)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [3700/6250]  eta: 0:11:24  lr: 0.000003  grad: 0.2067 (0.3669)  loss: 0.7871 (0.7839)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [90]  [3800/6250]  eta: 0:10:57  lr: 0.000003  grad: 0.1959 (0.3693)  loss: 0.7856 (0.7839)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [90]  [3900/6250]  eta: 0:10:30  lr: 0.000003  grad: 0.2237 (0.3685)  loss: 0.7719 (0.7838)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [90]  [4000/6250]  eta: 0:10:04  lr: 0.000003  grad: 0.2183 (0.3670)  loss: 0.7850 (0.7838)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [90]  [4100/6250]  eta: 0:09:37  lr: 0.000003  grad: 0.2112 (0.3875)  loss: 0.7854 (0.7837)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [90]  [4200/6250]  eta: 0:09:10  lr: 0.000003  grad: 0.2035 (0.3854)  loss: 0.7787 (0.7837)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [4300/6250]  eta: 0:08:43  lr: 0.000003  grad: 0.2018 (0.3839)  loss: 0.7899 (0.7836)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [90]  [4400/6250]  eta: 0:08:16  lr: 0.000003  grad: 0.2176 (0.3893)  loss: 0.7807 (0.7835)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [90]  [4500/6250]  eta: 0:07:49  lr: 0.000003  grad: 0.2050 (0.3948)  loss: 0.7842 (0.7835)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [90]  [4600/6250]  eta: 0:07:22  lr: 0.000003  grad: 0.2310 (0.3943)  loss: 0.7780 (0.7835)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [90]  [4700/6250]  eta: 0:06:55  lr: 0.000003  grad: 0.2118 (0.3913)  loss: 0.7874 (0.7834)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [90]  [4800/6250]  eta: 0:06:28  lr: 0.000003  grad: 0.2191 (0.3931)  loss: 0.7796 (0.7834)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [90]  [4900/6250]  eta: 0:06:02  lr: 0.000003  grad: 0.2183 (0.3934)  loss: 0.7774 (0.7834)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [90]  [5000/6250]  eta: 0:05:35  lr: 0.000003  grad: 0.2106 (0.3936)  loss: 0.7836 (0.7835)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [90]  [5100/6250]  eta: 0:05:08  lr: 0.000003  grad: 0.2032 (0.3985)  loss: 0.7783 (0.7834)  time: 0.2658  data: 0.0001  max mem: 26157
Train: [90]  [5200/6250]  eta: 0:04:41  lr: 0.000003  grad: 0.2043 (0.4027)  loss: 0.7859 (0.7835)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [90]  [5300/6250]  eta: 0:04:14  lr: 0.000003  grad: 0.2050 (0.4040)  loss: 0.7895 (0.7834)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [5400/6250]  eta: 0:03:47  lr: 0.000003  grad: 0.2093 (0.4048)  loss: 0.7830 (0.7834)  time: 0.2654  data: 0.0001  max mem: 26157
Train: [90]  [5500/6250]  eta: 0:03:20  lr: 0.000003  grad: 0.2078 (0.4041)  loss: 0.7764 (0.7833)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [90]  [5600/6250]  eta: 0:02:54  lr: 0.000003  grad: 0.2059 (0.4028)  loss: 0.7847 (0.7833)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [5700/6250]  eta: 0:02:27  lr: 0.000003  grad: 0.2275 (0.4017)  loss: 0.7778 (0.7832)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [90]  [5800/6250]  eta: 0:02:00  lr: 0.000003  grad: 0.2124 (0.4035)  loss: 0.7756 (0.7831)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [90]  [5900/6250]  eta: 0:01:33  lr: 0.000003  grad: 0.2056 (0.4033)  loss: 0.7858 (0.7831)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [90]  [6000/6250]  eta: 0:01:06  lr: 0.000003  grad: 0.2301 (0.4034)  loss: 0.7870 (0.7831)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [90]  [6100/6250]  eta: 0:00:40  lr: 0.000003  grad: 0.2105 (0.4030)  loss: 0.7756 (0.7830)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [90]  [6200/6250]  eta: 0:00:13  lr: 0.000003  grad: 0.2102 (0.4009)  loss: 0.7814 (0.7829)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [90]  [6249/6250]  eta: 0:00:00  lr: 0.000003  grad: 0.2176 (0.4013)  loss: 0.7829 (0.7829)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [90] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000003  grad: 0.2176 (0.4013)  loss: 0.7829 (0.7829)
Eval (hcp-train-subset): [90]  [ 0/62]  eta: 0:04:10  loss: 0.7963 (0.7963)  time: 4.0411  data: 3.9586  max mem: 26157
Eval (hcp-train-subset): [90]  [61/62]  eta: 0:00:00  loss: 0.7810 (0.7881)  time: 0.0826  data: 0.0001  max mem: 26157
Eval (hcp-train-subset): [90] Total time: 0:00:10 (0.1726 s / it)
Averaged stats (hcp-train-subset): loss: 0.7810 (0.7881)
Making plots (hcp-train-subset): example=17
Eval (hcp-val): [90]  [ 0/62]  eta: 0:03:11  loss: 0.8200 (0.8200)  time: 3.0891  data: 2.9740  max mem: 26157
Eval (hcp-val): [90]  [61/62]  eta: 0:00:00  loss: 0.8216 (0.8233)  time: 0.1174  data: 0.0349  max mem: 26157
Eval (hcp-val): [90] Total time: 0:00:11 (0.1876 s / it)
Averaged stats (hcp-val): loss: 0.8216 (0.8233)
Making plots (hcp-val): example=40
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [91]  [   0/6250]  eta: 9:38:52  lr: 0.000003  grad: 0.5643 (0.5643)  loss: 0.7245 (0.7245)  time: 5.5571  data: 5.2755  max mem: 26157
Train: [91]  [ 100/6250]  eta: 0:32:48  lr: 0.000003  grad: 0.2763 (0.3980)  loss: 0.7876 (0.7926)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [91]  [ 200/6250]  eta: 0:29:36  lr: 0.000003  grad: 0.2734 (0.5028)  loss: 0.7844 (0.7876)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [91]  [ 300/6250]  eta: 0:28:14  lr: 0.000003  grad: 0.2239 (0.4351)  loss: 0.7872 (0.7849)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [91]  [ 400/6250]  eta: 0:27:20  lr: 0.000003  grad: 0.2265 (0.4643)  loss: 0.7776 (0.7838)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [91]  [ 500/6250]  eta: 0:26:36  lr: 0.000003  grad: 0.2670 (0.4393)  loss: 0.7800 (0.7832)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [91]  [ 600/6250]  eta: 0:25:58  lr: 0.000003  grad: 0.2418 (0.4635)  loss: 0.7699 (0.7826)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [91]  [ 700/6250]  eta: 0:25:24  lr: 0.000003  grad: 0.2272 (0.4573)  loss: 0.7903 (0.7829)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [91]  [ 800/6250]  eta: 0:24:51  lr: 0.000003  grad: 0.2212 (0.4393)  loss: 0.7877 (0.7834)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [91]  [ 900/6250]  eta: 0:24:20  lr: 0.000003  grad: 0.2117 (0.4458)  loss: 0.7851 (0.7836)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [91]  [1000/6250]  eta: 0:23:50  lr: 0.000003  grad: 0.2333 (0.4394)  loss: 0.7936 (0.7838)  time: 0.2717  data: 0.0002  max mem: 26157
Train: [91]  [1100/6250]  eta: 0:23:20  lr: 0.000003  grad: 0.2091 (0.4475)  loss: 0.7845 (0.7840)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [91]  [1200/6250]  eta: 0:22:51  lr: 0.000003  grad: 0.2230 (0.4516)  loss: 0.7816 (0.7841)  time: 0.2659  data: 0.0002  max mem: 26157
Train: [91]  [1300/6250]  eta: 0:22:22  lr: 0.000003  grad: 0.2099 (0.4393)  loss: 0.7899 (0.7842)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [91]  [1400/6250]  eta: 0:21:53  lr: 0.000003  grad: 0.2025 (0.4262)  loss: 0.7907 (0.7843)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [91]  [1500/6250]  eta: 0:21:25  lr: 0.000003  grad: 0.2127 (0.4337)  loss: 0.7843 (0.7843)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [91]  [1600/6250]  eta: 0:20:57  lr: 0.000003  grad: 0.2111 (0.4254)  loss: 0.7788 (0.7843)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [91]  [1700/6250]  eta: 0:20:29  lr: 0.000003  grad: 0.2065 (0.4287)  loss: 0.7874 (0.7843)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [91]  [1800/6250]  eta: 0:20:01  lr: 0.000003  grad: 0.2078 (0.4256)  loss: 0.7785 (0.7841)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [91]  [1900/6250]  eta: 0:19:34  lr: 0.000003  grad: 0.2267 (0.4210)  loss: 0.7801 (0.7841)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [91]  [2000/6250]  eta: 0:19:06  lr: 0.000003  grad: 0.2150 (0.4187)  loss: 0.7827 (0.7841)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [91]  [2100/6250]  eta: 0:18:38  lr: 0.000003  grad: 0.2064 (0.4116)  loss: 0.7891 (0.7839)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [91]  [2200/6250]  eta: 0:18:11  lr: 0.000003  grad: 0.2241 (0.4049)  loss: 0.7861 (0.7838)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [91]  [2300/6250]  eta: 0:17:44  lr: 0.000003  grad: 0.2128 (0.4040)  loss: 0.7800 (0.7836)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [91]  [2400/6250]  eta: 0:17:16  lr: 0.000003  grad: 0.2244 (0.4004)  loss: 0.7740 (0.7834)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [91]  [2500/6250]  eta: 0:16:49  lr: 0.000003  grad: 0.2021 (0.4019)  loss: 0.7844 (0.7833)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [91]  [2600/6250]  eta: 0:16:22  lr: 0.000003  grad: 0.2216 (0.4061)  loss: 0.7785 (0.7832)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [91]  [2700/6250]  eta: 0:15:55  lr: 0.000002  grad: 0.2224 (0.4088)  loss: 0.7811 (0.7830)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [91]  [2800/6250]  eta: 0:15:27  lr: 0.000002  grad: 0.2105 (0.4102)  loss: 0.7832 (0.7830)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [91]  [2900/6250]  eta: 0:15:00  lr: 0.000002  grad: 0.2129 (0.4101)  loss: 0.7808 (0.7828)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [91]  [3000/6250]  eta: 0:14:33  lr: 0.000002  grad: 0.2136 (0.4077)  loss: 0.7819 (0.7827)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [91]  [3100/6250]  eta: 0:14:06  lr: 0.000002  grad: 0.2200 (0.4050)  loss: 0.7743 (0.7825)  time: 0.2659  data: 0.0002  max mem: 26157
Train: [91]  [3200/6250]  eta: 0:13:39  lr: 0.000002  grad: 0.2156 (0.4118)  loss: 0.7767 (0.7824)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [91]  [3300/6250]  eta: 0:13:12  lr: 0.000002  grad: 0.2168 (0.4082)  loss: 0.7762 (0.7823)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [91]  [3400/6250]  eta: 0:12:45  lr: 0.000002  grad: 0.2146 (0.4082)  loss: 0.7758 (0.7822)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [91]  [3500/6250]  eta: 0:12:18  lr: 0.000002  grad: 0.2165 (0.4080)  loss: 0.7750 (0.7822)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [91]  [3600/6250]  eta: 0:11:51  lr: 0.000002  grad: 0.1965 (0.4036)  loss: 0.7827 (0.7821)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [91]  [3700/6250]  eta: 0:11:24  lr: 0.000002  grad: 0.2288 (0.4035)  loss: 0.7730 (0.7820)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [91]  [3800/6250]  eta: 0:10:57  lr: 0.000002  grad: 0.2064 (0.4011)  loss: 0.7795 (0.7820)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [91]  [3900/6250]  eta: 0:10:30  lr: 0.000002  grad: 0.2059 (0.4025)  loss: 0.7720 (0.7819)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [91]  [4000/6250]  eta: 0:10:03  lr: 0.000002  grad: 0.2102 (0.3995)  loss: 0.7811 (0.7818)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [91]  [4100/6250]  eta: 0:09:36  lr: 0.000002  grad: 0.2040 (0.3973)  loss: 0.7846 (0.7818)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [91]  [4200/6250]  eta: 0:09:10  lr: 0.000002  grad: 0.2352 (0.3949)  loss: 0.7762 (0.7817)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [91]  [4300/6250]  eta: 0:08:43  lr: 0.000002  grad: 0.2151 (0.3955)  loss: 0.7867 (0.7817)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [91]  [4400/6250]  eta: 0:08:16  lr: 0.000002  grad: 0.2124 (0.3970)  loss: 0.7857 (0.7818)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [91]  [4500/6250]  eta: 0:07:49  lr: 0.000002  grad: 0.2061 (0.3983)  loss: 0.7819 (0.7817)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [91]  [4600/6250]  eta: 0:07:22  lr: 0.000002  grad: 0.2103 (0.3970)  loss: 0.7829 (0.7817)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [91]  [4700/6250]  eta: 0:06:55  lr: 0.000002  grad: 0.2055 (0.3977)  loss: 0.7780 (0.7817)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [91]  [4800/6250]  eta: 0:06:28  lr: 0.000002  grad: 0.2050 (0.4023)  loss: 0.7870 (0.7818)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [91]  [4900/6250]  eta: 0:06:01  lr: 0.000002  grad: 0.2110 (0.4039)  loss: 0.7843 (0.7818)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [91]  [5000/6250]  eta: 0:05:35  lr: 0.000002  grad: 0.2011 (0.4012)  loss: 0.7901 (0.7819)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [91]  [5100/6250]  eta: 0:05:08  lr: 0.000002  grad: 0.2069 (0.4064)  loss: 0.7835 (0.7820)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [91]  [5200/6250]  eta: 0:04:41  lr: 0.000002  grad: 0.2000 (0.4057)  loss: 0.7834 (0.7821)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [91]  [5300/6250]  eta: 0:04:14  lr: 0.000002  grad: 0.2127 (0.4045)  loss: 0.7800 (0.7822)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [91]  [5400/6250]  eta: 0:03:47  lr: 0.000002  grad: 0.2181 (0.4043)  loss: 0.7819 (0.7823)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [91]  [5500/6250]  eta: 0:03:21  lr: 0.000002  grad: 0.2105 (0.4022)  loss: 0.7782 (0.7823)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [91]  [5600/6250]  eta: 0:02:54  lr: 0.000002  grad: 0.2218 (0.4032)  loss: 0.7755 (0.7822)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [91]  [5700/6250]  eta: 0:02:27  lr: 0.000002  grad: 0.2138 (0.4021)  loss: 0.7804 (0.7822)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [91]  [5800/6250]  eta: 0:02:00  lr: 0.000002  grad: 0.2102 (0.4018)  loss: 0.7836 (0.7822)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [91]  [5900/6250]  eta: 0:01:33  lr: 0.000002  grad: 0.2079 (0.4005)  loss: 0.7837 (0.7822)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [91]  [6000/6250]  eta: 0:01:06  lr: 0.000002  grad: 0.2241 (0.3996)  loss: 0.7707 (0.7822)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [91]  [6100/6250]  eta: 0:00:40  lr: 0.000002  grad: 0.2035 (0.4004)  loss: 0.7835 (0.7822)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [91]  [6200/6250]  eta: 0:00:13  lr: 0.000002  grad: 0.2047 (0.4009)  loss: 0.7879 (0.7822)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [91]  [6249/6250]  eta: 0:00:00  lr: 0.000002  grad: 0.1997 (0.4002)  loss: 0.7841 (0.7822)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [91] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000002  grad: 0.1997 (0.4002)  loss: 0.7841 (0.7822)
Eval (hcp-train-subset): [91]  [ 0/62]  eta: 0:02:55  loss: 0.7952 (0.7952)  time: 2.8268  data: 2.7286  max mem: 26157
Eval (hcp-train-subset): [91]  [61/62]  eta: 0:00:00  loss: 0.7823 (0.7869)  time: 0.0824  data: 0.0001  max mem: 26157
Eval (hcp-train-subset): [91] Total time: 0:00:10 (0.1667 s / it)
Averaged stats (hcp-train-subset): loss: 0.7823 (0.7869)
Making plots (hcp-train-subset): example=5
Eval (hcp-val): [91]  [ 0/62]  eta: 0:03:11  loss: 0.8214 (0.8214)  time: 3.0829  data: 2.9857  max mem: 26157
Eval (hcp-val): [91]  [61/62]  eta: 0:00:00  loss: 0.8222 (0.8229)  time: 0.0959  data: 0.0120  max mem: 26157
Eval (hcp-val): [91] Total time: 0:00:10 (0.1660 s / it)
Averaged stats (hcp-val): loss: 0.8222 (0.8229)
Making plots (hcp-val): example=57
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [92]  [   0/6250]  eta: 8:41:04  lr: 0.000002  grad: 0.2067 (0.2067)  loss: 0.8023 (0.8023)  time: 5.0024  data: 4.7198  max mem: 26157
Train: [92]  [ 100/6250]  eta: 0:32:19  lr: 0.000002  grad: 0.2790 (0.3754)  loss: 0.7881 (0.8006)  time: 0.2691  data: 0.0001  max mem: 26157
Train: [92]  [ 200/6250]  eta: 0:29:26  lr: 0.000002  grad: 0.2668 (0.3669)  loss: 0.7758 (0.7889)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [92]  [ 300/6250]  eta: 0:28:11  lr: 0.000002  grad: 0.2547 (0.4051)  loss: 0.7848 (0.7856)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [92]  [ 400/6250]  eta: 0:27:19  lr: 0.000002  grad: 0.2117 (0.3869)  loss: 0.7843 (0.7843)  time: 0.2700  data: 0.0002  max mem: 26157
Train: [92]  [ 500/6250]  eta: 0:26:36  lr: 0.000002  grad: 0.2383 (0.3874)  loss: 0.7826 (0.7835)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [92]  [ 600/6250]  eta: 0:25:58  lr: 0.000002  grad: 0.2334 (0.3932)  loss: 0.7849 (0.7838)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [92]  [ 700/6250]  eta: 0:25:24  lr: 0.000002  grad: 0.2146 (0.4264)  loss: 0.7755 (0.7840)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [92]  [ 800/6250]  eta: 0:24:51  lr: 0.000002  grad: 0.2179 (0.4191)  loss: 0.7857 (0.7843)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [92]  [ 900/6250]  eta: 0:24:20  lr: 0.000002  grad: 0.2187 (0.4092)  loss: 0.7903 (0.7846)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [92]  [1000/6250]  eta: 0:23:50  lr: 0.000002  grad: 0.2331 (0.4099)  loss: 0.7818 (0.7847)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [92]  [1100/6250]  eta: 0:23:20  lr: 0.000002  grad: 0.2265 (0.4178)  loss: 0.7782 (0.7847)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [92]  [1200/6250]  eta: 0:22:51  lr: 0.000002  grad: 0.2180 (0.4051)  loss: 0.7843 (0.7846)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [92]  [1300/6250]  eta: 0:22:22  lr: 0.000002  grad: 0.2210 (0.3986)  loss: 0.7879 (0.7843)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [92]  [1400/6250]  eta: 0:21:53  lr: 0.000002  grad: 0.2171 (0.4028)  loss: 0.7828 (0.7842)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [92]  [1500/6250]  eta: 0:21:25  lr: 0.000002  grad: 0.2214 (0.3938)  loss: 0.7759 (0.7840)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [92]  [1600/6250]  eta: 0:20:57  lr: 0.000002  grad: 0.2164 (0.4044)  loss: 0.7840 (0.7839)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [92]  [1700/6250]  eta: 0:20:29  lr: 0.000002  grad: 0.2091 (0.3980)  loss: 0.7756 (0.7836)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [92]  [1800/6250]  eta: 0:20:02  lr: 0.000002  grad: 0.2225 (0.4060)  loss: 0.7774 (0.7834)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [92]  [1900/6250]  eta: 0:19:34  lr: 0.000002  grad: 0.2289 (0.4072)  loss: 0.7821 (0.7835)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [92]  [2000/6250]  eta: 0:19:06  lr: 0.000002  grad: 0.2143 (0.4035)  loss: 0.7850 (0.7835)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [92]  [2100/6250]  eta: 0:18:39  lr: 0.000002  grad: 0.2068 (0.4031)  loss: 0.7858 (0.7836)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [92]  [2200/6250]  eta: 0:18:12  lr: 0.000002  grad: 0.2186 (0.4090)  loss: 0.7794 (0.7836)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [92]  [2300/6250]  eta: 0:17:44  lr: 0.000002  grad: 0.2150 (0.4137)  loss: 0.7802 (0.7835)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [92]  [2400/6250]  eta: 0:17:17  lr: 0.000002  grad: 0.1978 (0.4071)  loss: 0.7813 (0.7835)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [92]  [2500/6250]  eta: 0:16:49  lr: 0.000002  grad: 0.2209 (0.4066)  loss: 0.7800 (0.7835)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [92]  [2600/6250]  eta: 0:16:22  lr: 0.000002  grad: 0.2227 (0.4011)  loss: 0.7804 (0.7834)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [92]  [2700/6250]  eta: 0:15:55  lr: 0.000002  grad: 0.2115 (0.3970)  loss: 0.7782 (0.7834)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [92]  [2800/6250]  eta: 0:15:28  lr: 0.000002  grad: 0.2204 (0.3987)  loss: 0.7843 (0.7833)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [92]  [2900/6250]  eta: 0:15:01  lr: 0.000002  grad: 0.2140 (0.3964)  loss: 0.7837 (0.7833)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [92]  [3000/6250]  eta: 0:14:34  lr: 0.000002  grad: 0.2131 (0.3936)  loss: 0.7811 (0.7832)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [92]  [3100/6250]  eta: 0:14:06  lr: 0.000002  grad: 0.2166 (0.3883)  loss: 0.7782 (0.7833)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [92]  [3200/6250]  eta: 0:13:39  lr: 0.000002  grad: 0.2308 (0.3888)  loss: 0.7812 (0.7832)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [92]  [3300/6250]  eta: 0:13:12  lr: 0.000002  grad: 0.2278 (0.3900)  loss: 0.7816 (0.7832)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [92]  [3400/6250]  eta: 0:12:45  lr: 0.000002  grad: 0.2316 (0.3874)  loss: 0.7811 (0.7832)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [92]  [3500/6250]  eta: 0:12:18  lr: 0.000002  grad: 0.2219 (0.3833)  loss: 0.7836 (0.7832)  time: 0.2712  data: 0.0002  max mem: 26157
Train: [92]  [3600/6250]  eta: 0:11:52  lr: 0.000002  grad: 0.2117 (0.3913)  loss: 0.7840 (0.7832)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [92]  [3700/6250]  eta: 0:11:25  lr: 0.000002  grad: 0.2234 (0.3957)  loss: 0.7799 (0.7832)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [92]  [3800/6250]  eta: 0:10:58  lr: 0.000002  grad: 0.2301 (0.3974)  loss: 0.7757 (0.7830)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [92]  [3900/6250]  eta: 0:10:31  lr: 0.000002  grad: 0.2222 (0.3981)  loss: 0.7790 (0.7830)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [92]  [4000/6250]  eta: 0:10:04  lr: 0.000002  grad: 0.2181 (0.4070)  loss: 0.7872 (0.7830)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [92]  [4100/6250]  eta: 0:09:37  lr: 0.000002  grad: 0.2250 (0.4081)  loss: 0.7875 (0.7829)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [92]  [4200/6250]  eta: 0:09:10  lr: 0.000002  grad: 0.2184 (0.4134)  loss: 0.7786 (0.7828)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [92]  [4300/6250]  eta: 0:08:43  lr: 0.000002  grad: 0.2303 (0.4111)  loss: 0.7773 (0.7827)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [92]  [4400/6250]  eta: 0:08:16  lr: 0.000002  grad: 0.2370 (0.4107)  loss: 0.7738 (0.7827)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [92]  [4500/6250]  eta: 0:07:49  lr: 0.000002  grad: 0.2096 (0.4108)  loss: 0.7837 (0.7826)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [92]  [4600/6250]  eta: 0:07:22  lr: 0.000002  grad: 0.2210 (0.4120)  loss: 0.7794 (0.7826)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [92]  [4700/6250]  eta: 0:06:55  lr: 0.000002  grad: 0.2278 (0.4102)  loss: 0.7893 (0.7826)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [92]  [4800/6250]  eta: 0:06:28  lr: 0.000002  grad: 0.2212 (0.4117)  loss: 0.7798 (0.7826)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [92]  [4900/6250]  eta: 0:06:02  lr: 0.000002  grad: 0.2194 (0.4115)  loss: 0.7807 (0.7826)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [92]  [5000/6250]  eta: 0:05:35  lr: 0.000002  grad: 0.2212 (0.4098)  loss: 0.7768 (0.7825)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [92]  [5100/6250]  eta: 0:05:08  lr: 0.000002  grad: 0.2326 (0.4071)  loss: 0.7862 (0.7825)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [92]  [5200/6250]  eta: 0:04:41  lr: 0.000002  grad: 0.2057 (0.4060)  loss: 0.7870 (0.7825)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [92]  [5300/6250]  eta: 0:04:14  lr: 0.000002  grad: 0.2247 (0.4059)  loss: 0.7784 (0.7824)  time: 0.2687  data: 0.0001  max mem: 26157
Train: [92]  [5400/6250]  eta: 0:03:47  lr: 0.000002  grad: 0.2206 (0.4054)  loss: 0.7762 (0.7824)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [92]  [5500/6250]  eta: 0:03:21  lr: 0.000002  grad: 0.1997 (0.4035)  loss: 0.7754 (0.7823)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [92]  [5600/6250]  eta: 0:02:54  lr: 0.000002  grad: 0.2152 (0.4038)  loss: 0.7761 (0.7823)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [92]  [5700/6250]  eta: 0:02:27  lr: 0.000002  grad: 0.2179 (0.4031)  loss: 0.7769 (0.7822)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [92]  [5800/6250]  eta: 0:02:00  lr: 0.000002  grad: 0.2139 (0.4044)  loss: 0.7820 (0.7822)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [92]  [5900/6250]  eta: 0:01:33  lr: 0.000002  grad: 0.2139 (0.4014)  loss: 0.7780 (0.7821)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [92]  [6000/6250]  eta: 0:01:06  lr: 0.000002  grad: 0.2171 (0.4006)  loss: 0.7752 (0.7820)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [92]  [6100/6250]  eta: 0:00:40  lr: 0.000002  grad: 0.2120 (0.3986)  loss: 0.7833 (0.7819)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [92]  [6200/6250]  eta: 0:00:13  lr: 0.000002  grad: 0.2193 (0.3977)  loss: 0.7738 (0.7818)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [92]  [6249/6250]  eta: 0:00:00  lr: 0.000002  grad: 0.2134 (0.3969)  loss: 0.7722 (0.7817)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [92] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000002  grad: 0.2134 (0.3969)  loss: 0.7722 (0.7817)
Eval (hcp-train-subset): [92]  [ 0/62]  eta: 0:04:46  loss: 0.7920 (0.7920)  time: 4.6216  data: 4.5383  max mem: 26157
Eval (hcp-train-subset): [92]  [61/62]  eta: 0:00:00  loss: 0.7810 (0.7857)  time: 0.1156  data: 0.0331  max mem: 26157
Eval (hcp-train-subset): [92] Total time: 0:00:11 (0.1866 s / it)
Averaged stats (hcp-train-subset): loss: 0.7810 (0.7857)
Making plots (hcp-train-subset): example=30
Eval (hcp-val): [92]  [ 0/62]  eta: 0:04:51  loss: 0.8167 (0.8167)  time: 4.7032  data: 4.6204  max mem: 26157
Eval (hcp-val): [92]  [61/62]  eta: 0:00:00  loss: 0.8224 (0.8229)  time: 0.0928  data: 0.0105  max mem: 26157
Eval (hcp-val): [92] Total time: 0:00:11 (0.1794 s / it)
Averaged stats (hcp-val): loss: 0.8224 (0.8229)
Making plots (hcp-val): example=33
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [93]  [   0/6250]  eta: 8:15:36  lr: 0.000002  grad: 0.2789 (0.2789)  loss: 0.8213 (0.8213)  time: 4.7579  data: 4.4867  max mem: 26157
Train: [93]  [ 100/6250]  eta: 0:31:59  lr: 0.000002  grad: 0.2803 (0.4082)  loss: 0.7969 (0.7862)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [93]  [ 200/6250]  eta: 0:29:12  lr: 0.000002  grad: 0.2351 (0.3594)  loss: 0.7835 (0.7850)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [93]  [ 300/6250]  eta: 0:27:59  lr: 0.000002  grad: 0.2300 (0.3674)  loss: 0.7884 (0.7840)  time: 0.2687  data: 0.0002  max mem: 26157
Train: [93]  [ 400/6250]  eta: 0:27:09  lr: 0.000002  grad: 0.2305 (0.3659)  loss: 0.7753 (0.7843)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [93]  [ 500/6250]  eta: 0:26:29  lr: 0.000002  grad: 0.2450 (0.3847)  loss: 0.7820 (0.7845)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [93]  [ 600/6250]  eta: 0:25:53  lr: 0.000002  grad: 0.2045 (0.3644)  loss: 0.7899 (0.7845)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [93]  [ 700/6250]  eta: 0:25:19  lr: 0.000002  grad: 0.2096 (0.3902)  loss: 0.7929 (0.7846)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [93]  [ 800/6250]  eta: 0:24:47  lr: 0.000002  grad: 0.2066 (0.3840)  loss: 0.7851 (0.7844)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [93]  [ 900/6250]  eta: 0:24:16  lr: 0.000002  grad: 0.2037 (0.3718)  loss: 0.7866 (0.7845)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [93]  [1000/6250]  eta: 0:23:46  lr: 0.000002  grad: 0.2141 (0.3766)  loss: 0.7879 (0.7845)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [93]  [1100/6250]  eta: 0:23:17  lr: 0.000002  grad: 0.2097 (0.3888)  loss: 0.7866 (0.7843)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [93]  [1200/6250]  eta: 0:22:48  lr: 0.000002  grad: 0.2106 (0.3815)  loss: 0.7803 (0.7841)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [93]  [1300/6250]  eta: 0:22:19  lr: 0.000002  grad: 0.2122 (0.3785)  loss: 0.7864 (0.7841)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [93]  [1400/6250]  eta: 0:21:51  lr: 0.000002  grad: 0.2274 (0.3865)  loss: 0.7866 (0.7840)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [93]  [1500/6250]  eta: 0:21:22  lr: 0.000002  grad: 0.1934 (0.4013)  loss: 0.7786 (0.7841)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [93]  [1600/6250]  eta: 0:20:55  lr: 0.000002  grad: 0.2106 (0.3946)  loss: 0.7853 (0.7841)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [93]  [1700/6250]  eta: 0:20:27  lr: 0.000002  grad: 0.2073 (0.3868)  loss: 0.7846 (0.7841)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [93]  [1800/6250]  eta: 0:19:59  lr: 0.000002  grad: 0.2104 (0.3818)  loss: 0.7789 (0.7841)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [93]  [1900/6250]  eta: 0:19:32  lr: 0.000002  grad: 0.2235 (0.3791)  loss: 0.7805 (0.7840)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [93]  [2000/6250]  eta: 0:19:04  lr: 0.000002  grad: 0.2082 (0.3776)  loss: 0.7868 (0.7840)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [93]  [2100/6250]  eta: 0:18:37  lr: 0.000002  grad: 0.2169 (0.3748)  loss: 0.7819 (0.7837)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [93]  [2200/6250]  eta: 0:18:09  lr: 0.000002  grad: 0.2060 (0.3750)  loss: 0.7774 (0.7837)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [93]  [2300/6250]  eta: 0:17:42  lr: 0.000001  grad: 0.2071 (0.3742)  loss: 0.7829 (0.7836)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [93]  [2400/6250]  eta: 0:17:15  lr: 0.000001  grad: 0.2292 (0.3717)  loss: 0.7796 (0.7835)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [93]  [2500/6250]  eta: 0:16:48  lr: 0.000001  grad: 0.2155 (0.3715)  loss: 0.7859 (0.7836)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [93]  [2600/6250]  eta: 0:16:21  lr: 0.000001  grad: 0.1981 (0.3698)  loss: 0.7842 (0.7835)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [93]  [2700/6250]  eta: 0:15:53  lr: 0.000001  grad: 0.2026 (0.3657)  loss: 0.7775 (0.7834)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [93]  [2800/6250]  eta: 0:15:26  lr: 0.000001  grad: 0.2125 (0.3698)  loss: 0.7823 (0.7833)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [93]  [2900/6250]  eta: 0:14:59  lr: 0.000001  grad: 0.2156 (0.3677)  loss: 0.7769 (0.7831)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [93]  [3000/6250]  eta: 0:14:32  lr: 0.000001  grad: 0.2159 (0.3656)  loss: 0.7895 (0.7831)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [93]  [3100/6250]  eta: 0:14:05  lr: 0.000001  grad: 0.2264 (0.3719)  loss: 0.7782 (0.7829)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [93]  [3200/6250]  eta: 0:13:38  lr: 0.000001  grad: 0.2092 (0.3723)  loss: 0.7806 (0.7828)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [93]  [3300/6250]  eta: 0:13:11  lr: 0.000001  grad: 0.2188 (0.3696)  loss: 0.7773 (0.7828)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [93]  [3400/6250]  eta: 0:12:44  lr: 0.000001  grad: 0.2381 (0.3740)  loss: 0.7813 (0.7827)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [93]  [3500/6250]  eta: 0:12:17  lr: 0.000001  grad: 0.2200 (0.3711)  loss: 0.7751 (0.7826)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [93]  [3600/6250]  eta: 0:11:50  lr: 0.000001  grad: 0.2126 (0.3714)  loss: 0.7790 (0.7825)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [93]  [3700/6250]  eta: 0:11:23  lr: 0.000001  grad: 0.2067 (0.3737)  loss: 0.7838 (0.7825)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [93]  [3800/6250]  eta: 0:10:57  lr: 0.000001  grad: 0.2212 (0.3779)  loss: 0.7741 (0.7823)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [93]  [3900/6250]  eta: 0:10:30  lr: 0.000001  grad: 0.2234 (0.3808)  loss: 0.7697 (0.7822)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [93]  [4000/6250]  eta: 0:10:03  lr: 0.000001  grad: 0.2278 (0.3822)  loss: 0.7765 (0.7821)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [93]  [4100/6250]  eta: 0:09:36  lr: 0.000001  grad: 0.2038 (0.3849)  loss: 0.7854 (0.7821)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [93]  [4200/6250]  eta: 0:09:09  lr: 0.000001  grad: 0.2143 (0.3852)  loss: 0.7859 (0.7821)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [93]  [4300/6250]  eta: 0:08:42  lr: 0.000001  grad: 0.2211 (0.3835)  loss: 0.7817 (0.7821)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [93]  [4400/6250]  eta: 0:08:15  lr: 0.000001  grad: 0.2177 (0.3897)  loss: 0.7841 (0.7821)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [93]  [4500/6250]  eta: 0:07:48  lr: 0.000001  grad: 0.2195 (0.3875)  loss: 0.7820 (0.7821)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [93]  [4600/6250]  eta: 0:07:22  lr: 0.000001  grad: 0.2151 (0.3981)  loss: 0.7899 (0.7822)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [93]  [4700/6250]  eta: 0:06:55  lr: 0.000001  grad: 0.2324 (0.3989)  loss: 0.7836 (0.7822)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [93]  [4800/6250]  eta: 0:06:28  lr: 0.000001  grad: 0.2109 (0.4035)  loss: 0.7808 (0.7823)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [93]  [4900/6250]  eta: 0:06:01  lr: 0.000001  grad: 0.2040 (0.4005)  loss: 0.7857 (0.7823)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [93]  [5000/6250]  eta: 0:05:34  lr: 0.000001  grad: 0.2291 (0.3978)  loss: 0.7809 (0.7824)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [93]  [5100/6250]  eta: 0:05:07  lr: 0.000001  grad: 0.2137 (0.3964)  loss: 0.7876 (0.7824)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [93]  [5200/6250]  eta: 0:04:41  lr: 0.000001  grad: 0.2043 (0.3951)  loss: 0.7791 (0.7824)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [93]  [5300/6250]  eta: 0:04:14  lr: 0.000001  grad: 0.2185 (0.3976)  loss: 0.7694 (0.7824)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [93]  [5400/6250]  eta: 0:03:47  lr: 0.000001  grad: 0.2082 (0.3991)  loss: 0.7849 (0.7824)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [93]  [5500/6250]  eta: 0:03:20  lr: 0.000001  grad: 0.2233 (0.3999)  loss: 0.7765 (0.7824)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [93]  [5600/6250]  eta: 0:02:54  lr: 0.000001  grad: 0.2006 (0.4020)  loss: 0.7846 (0.7824)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [93]  [5700/6250]  eta: 0:02:27  lr: 0.000001  grad: 0.2190 (0.4009)  loss: 0.7807 (0.7824)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [93]  [5800/6250]  eta: 0:02:00  lr: 0.000001  grad: 0.2156 (0.3998)  loss: 0.7828 (0.7825)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [93]  [5900/6250]  eta: 0:01:33  lr: 0.000001  grad: 0.2170 (0.3990)  loss: 0.7880 (0.7824)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [93]  [6000/6250]  eta: 0:01:06  lr: 0.000001  grad: 0.1997 (0.3992)  loss: 0.7887 (0.7825)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [93]  [6100/6250]  eta: 0:00:40  lr: 0.000001  grad: 0.2133 (0.3998)  loss: 0.7837 (0.7825)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [93]  [6200/6250]  eta: 0:00:13  lr: 0.000001  grad: 0.2205 (0.3998)  loss: 0.7888 (0.7826)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [93]  [6249/6250]  eta: 0:00:00  lr: 0.000001  grad: 0.2110 (0.4008)  loss: 0.7904 (0.7826)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [93] Total time: 0:27:55 (0.2682 s / it)
Averaged stats: lr: 0.000001  grad: 0.2110 (0.4008)  loss: 0.7904 (0.7826)
Eval (hcp-train-subset): [93]  [ 0/62]  eta: 0:04:43  loss: 0.7886 (0.7886)  time: 4.5793  data: 4.4974  max mem: 26157
Eval (hcp-train-subset): [93]  [61/62]  eta: 0:00:00  loss: 0.7822 (0.7857)  time: 0.0940  data: 0.0115  max mem: 26157
Eval (hcp-train-subset): [93] Total time: 0:00:11 (0.1825 s / it)
Averaged stats (hcp-train-subset): loss: 0.7822 (0.7857)
Making plots (hcp-train-subset): example=42
Eval (hcp-val): [93]  [ 0/62]  eta: 0:04:18  loss: 0.8168 (0.8168)  time: 4.1761  data: 4.0930  max mem: 26157
Eval (hcp-val): [93]  [61/62]  eta: 0:00:00  loss: 0.8212 (0.8224)  time: 0.0927  data: 0.0089  max mem: 26157
Eval (hcp-val): [93] Total time: 0:00:10 (0.1728 s / it)
Averaged stats (hcp-val): loss: 0.8212 (0.8224)
Making plots (hcp-val): example=62
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [94]  [   0/6250]  eta: 8:13:04  lr: 0.000001  grad: 0.2553 (0.2553)  loss: 0.8527 (0.8527)  time: 4.7336  data: 4.4623  max mem: 26157
Train: [94]  [ 100/6250]  eta: 0:32:00  lr: 0.000001  grad: 0.2942 (0.7642)  loss: 0.7746 (0.7850)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [94]  [ 200/6250]  eta: 0:29:14  lr: 0.000001  grad: 0.2428 (0.5738)  loss: 0.7677 (0.7823)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [94]  [ 300/6250]  eta: 0:28:00  lr: 0.000001  grad: 0.2611 (0.5599)  loss: 0.7799 (0.7810)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [94]  [ 400/6250]  eta: 0:27:10  lr: 0.000001  grad: 0.2303 (0.5422)  loss: 0.7803 (0.7802)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [94]  [ 500/6250]  eta: 0:26:29  lr: 0.000001  grad: 0.2428 (0.5086)  loss: 0.7837 (0.7803)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [94]  [ 600/6250]  eta: 0:25:53  lr: 0.000001  grad: 0.2389 (0.4812)  loss: 0.7827 (0.7808)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [94]  [ 700/6250]  eta: 0:25:19  lr: 0.000001  grad: 0.2241 (0.4695)  loss: 0.7883 (0.7815)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [94]  [ 800/6250]  eta: 0:24:47  lr: 0.000001  grad: 0.2180 (0.4487)  loss: 0.7836 (0.7819)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [94]  [ 900/6250]  eta: 0:24:17  lr: 0.000001  grad: 0.2198 (0.4435)  loss: 0.7858 (0.7824)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [94]  [1000/6250]  eta: 0:23:47  lr: 0.000001  grad: 0.2181 (0.4400)  loss: 0.7858 (0.7824)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [94]  [1100/6250]  eta: 0:23:17  lr: 0.000001  grad: 0.2224 (0.4253)  loss: 0.7836 (0.7825)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [94]  [1200/6250]  eta: 0:22:48  lr: 0.000001  grad: 0.2121 (0.4181)  loss: 0.7827 (0.7827)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [94]  [1300/6250]  eta: 0:22:20  lr: 0.000001  grad: 0.2245 (0.4282)  loss: 0.7850 (0.7827)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [94]  [1400/6250]  eta: 0:21:51  lr: 0.000001  grad: 0.2259 (0.4263)  loss: 0.7837 (0.7829)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [94]  [1500/6250]  eta: 0:21:23  lr: 0.000001  grad: 0.2127 (0.4454)  loss: 0.7892 (0.7831)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [94]  [1600/6250]  eta: 0:20:55  lr: 0.000001  grad: 0.2056 (0.4397)  loss: 0.7819 (0.7833)  time: 0.2667  data: 0.0002  max mem: 26157
Train: [94]  [1700/6250]  eta: 0:20:27  lr: 0.000001  grad: 0.2326 (0.4292)  loss: 0.7844 (0.7834)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [94]  [1800/6250]  eta: 0:20:00  lr: 0.000001  grad: 0.2110 (0.4291)  loss: 0.7808 (0.7834)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [94]  [1900/6250]  eta: 0:19:32  lr: 0.000001  grad: 0.2247 (0.4257)  loss: 0.7795 (0.7835)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [94]  [2000/6250]  eta: 0:19:05  lr: 0.000001  grad: 0.2303 (0.4229)  loss: 0.7854 (0.7836)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [94]  [2100/6250]  eta: 0:18:37  lr: 0.000001  grad: 0.2263 (0.4204)  loss: 0.7895 (0.7836)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [94]  [2200/6250]  eta: 0:18:10  lr: 0.000001  grad: 0.2210 (0.4192)  loss: 0.7845 (0.7837)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [94]  [2300/6250]  eta: 0:17:43  lr: 0.000001  grad: 0.2412 (0.4155)  loss: 0.7811 (0.7837)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [94]  [2400/6250]  eta: 0:17:15  lr: 0.000001  grad: 0.2420 (0.4134)  loss: 0.7787 (0.7837)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [94]  [2500/6250]  eta: 0:16:48  lr: 0.000001  grad: 0.2218 (0.4187)  loss: 0.7848 (0.7836)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [94]  [2600/6250]  eta: 0:16:21  lr: 0.000001  grad: 0.2401 (0.4140)  loss: 0.7802 (0.7834)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [94]  [2700/6250]  eta: 0:15:54  lr: 0.000001  grad: 0.2129 (0.4113)  loss: 0.7862 (0.7833)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [94]  [2800/6250]  eta: 0:15:27  lr: 0.000001  grad: 0.2364 (0.4153)  loss: 0.7791 (0.7833)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [94]  [2900/6250]  eta: 0:15:00  lr: 0.000001  grad: 0.2255 (0.4144)  loss: 0.7849 (0.7833)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [94]  [3000/6250]  eta: 0:14:33  lr: 0.000001  grad: 0.2246 (0.4127)  loss: 0.7789 (0.7832)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [94]  [3100/6250]  eta: 0:14:06  lr: 0.000001  grad: 0.2193 (0.4152)  loss: 0.7757 (0.7831)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [94]  [3200/6250]  eta: 0:13:39  lr: 0.000001  grad: 0.2154 (0.4248)  loss: 0.7801 (0.7830)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [94]  [3300/6250]  eta: 0:13:12  lr: 0.000001  grad: 0.2093 (0.4217)  loss: 0.7869 (0.7829)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [94]  [3400/6250]  eta: 0:12:45  lr: 0.000001  grad: 0.2094 (0.4203)  loss: 0.7873 (0.7828)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [94]  [3500/6250]  eta: 0:12:18  lr: 0.000001  grad: 0.2330 (0.4274)  loss: 0.7766 (0.7827)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [94]  [3600/6250]  eta: 0:11:51  lr: 0.000001  grad: 0.2158 (0.4230)  loss: 0.7766 (0.7825)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [94]  [3700/6250]  eta: 0:11:24  lr: 0.000001  grad: 0.2158 (0.4235)  loss: 0.7792 (0.7824)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [94]  [3800/6250]  eta: 0:10:57  lr: 0.000001  grad: 0.2226 (0.4259)  loss: 0.7721 (0.7822)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [94]  [3900/6250]  eta: 0:10:30  lr: 0.000001  grad: 0.2136 (0.4318)  loss: 0.7847 (0.7821)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [94]  [4000/6250]  eta: 0:10:03  lr: 0.000001  grad: 0.2177 (0.4303)  loss: 0.7837 (0.7820)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [94]  [4100/6250]  eta: 0:09:36  lr: 0.000001  grad: 0.2116 (0.4274)  loss: 0.7829 (0.7820)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [94]  [4200/6250]  eta: 0:09:09  lr: 0.000001  grad: 0.2124 (0.4243)  loss: 0.7776 (0.7819)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [94]  [4300/6250]  eta: 0:08:42  lr: 0.000001  grad: 0.2131 (0.4200)  loss: 0.7787 (0.7819)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [94]  [4400/6250]  eta: 0:08:15  lr: 0.000001  grad: 0.2081 (0.4191)  loss: 0.7844 (0.7819)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [94]  [4500/6250]  eta: 0:07:49  lr: 0.000001  grad: 0.2198 (0.4165)  loss: 0.7880 (0.7819)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [94]  [4600/6250]  eta: 0:07:22  lr: 0.000001  grad: 0.2064 (0.4147)  loss: 0.7855 (0.7820)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [94]  [4700/6250]  eta: 0:06:55  lr: 0.000001  grad: 0.2114 (0.4171)  loss: 0.7922 (0.7821)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [94]  [4800/6250]  eta: 0:06:28  lr: 0.000001  grad: 0.2142 (0.4141)  loss: 0.7886 (0.7822)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [94]  [4900/6250]  eta: 0:06:01  lr: 0.000001  grad: 0.2154 (0.4120)  loss: 0.7817 (0.7823)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [94]  [5000/6250]  eta: 0:05:34  lr: 0.000001  grad: 0.2122 (0.4118)  loss: 0.7953 (0.7824)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [94]  [5100/6250]  eta: 0:05:08  lr: 0.000001  grad: 0.2114 (0.4117)  loss: 0.7852 (0.7824)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [94]  [5200/6250]  eta: 0:04:41  lr: 0.000001  grad: 0.2029 (0.4125)  loss: 0.7883 (0.7825)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [94]  [5300/6250]  eta: 0:04:14  lr: 0.000001  grad: 0.2168 (0.4175)  loss: 0.7867 (0.7825)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [94]  [5400/6250]  eta: 0:03:47  lr: 0.000001  grad: 0.2240 (0.4155)  loss: 0.7782 (0.7826)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [94]  [5500/6250]  eta: 0:03:20  lr: 0.000001  grad: 0.2180 (0.4130)  loss: 0.7830 (0.7826)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [94]  [5600/6250]  eta: 0:02:54  lr: 0.000001  grad: 0.2078 (0.4138)  loss: 0.7848 (0.7827)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [94]  [5700/6250]  eta: 0:02:27  lr: 0.000001  grad: 0.2266 (0.4141)  loss: 0.7840 (0.7827)  time: 0.2655  data: 0.0001  max mem: 26157
Train: [94]  [5800/6250]  eta: 0:02:00  lr: 0.000001  grad: 0.2293 (0.4131)  loss: 0.7795 (0.7827)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [94]  [5900/6250]  eta: 0:01:33  lr: 0.000001  grad: 0.2135 (0.4182)  loss: 0.7784 (0.7827)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [94]  [6000/6250]  eta: 0:01:06  lr: 0.000001  grad: 0.2159 (0.4158)  loss: 0.7795 (0.7827)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [94]  [6100/6250]  eta: 0:00:40  lr: 0.000001  grad: 0.2192 (0.4186)  loss: 0.7871 (0.7827)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [94]  [6200/6250]  eta: 0:00:13  lr: 0.000001  grad: 0.2242 (0.4159)  loss: 0.7837 (0.7828)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [94]  [6249/6250]  eta: 0:00:00  lr: 0.000001  grad: 0.2154 (0.4155)  loss: 0.7906 (0.7828)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [94] Total time: 0:27:56 (0.2683 s / it)
Averaged stats: lr: 0.000001  grad: 0.2154 (0.4155)  loss: 0.7906 (0.7828)
Eval (hcp-train-subset): [94]  [ 0/62]  eta: 0:03:59  loss: 0.7858 (0.7858)  time: 3.8585  data: 3.7755  max mem: 26157
Eval (hcp-train-subset): [94]  [61/62]  eta: 0:00:00  loss: 0.7808 (0.7849)  time: 0.0824  data: 0.0001  max mem: 26157
Eval (hcp-train-subset): [94] Total time: 0:00:10 (0.1660 s / it)
Averaged stats (hcp-train-subset): loss: 0.7808 (0.7849)
Making plots (hcp-train-subset): example=39
Eval (hcp-val): [94]  [ 0/62]  eta: 0:04:15  loss: 0.8202 (0.8202)  time: 4.1283  data: 4.0455  max mem: 26157
Eval (hcp-val): [94]  [61/62]  eta: 0:00:00  loss: 0.8213 (0.8221)  time: 0.0883  data: 0.0059  max mem: 26157
Eval (hcp-val): [94] Total time: 0:00:10 (0.1684 s / it)
Averaged stats (hcp-val): loss: 0.8213 (0.8221)
Making plots (hcp-val): example=15
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [95]  [   0/6250]  eta: 9:14:53  lr: 0.000001  grad: 0.1914 (0.1914)  loss: 0.8216 (0.8216)  time: 5.3270  data: 5.0537  max mem: 26157
Train: [95]  [ 100/6250]  eta: 0:32:33  lr: 0.000001  grad: 0.2586 (0.3818)  loss: 0.7904 (0.7936)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [95]  [ 200/6250]  eta: 0:29:27  lr: 0.000001  grad: 0.2687 (0.4052)  loss: 0.7936 (0.7895)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [95]  [ 300/6250]  eta: 0:28:08  lr: 0.000001  grad: 0.2839 (0.3858)  loss: 0.7835 (0.7856)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [95]  [ 400/6250]  eta: 0:27:16  lr: 0.000001  grad: 0.2458 (0.3742)  loss: 0.7851 (0.7836)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [95]  [ 500/6250]  eta: 0:26:34  lr: 0.000001  grad: 0.2377 (0.4589)  loss: 0.7844 (0.7828)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [95]  [ 600/6250]  eta: 0:25:56  lr: 0.000001  grad: 0.2304 (0.4360)  loss: 0.7794 (0.7824)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [95]  [ 700/6250]  eta: 0:25:23  lr: 0.000001  grad: 0.2277 (0.4440)  loss: 0.7825 (0.7821)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [95]  [ 800/6250]  eta: 0:24:50  lr: 0.000001  grad: 0.2374 (0.4319)  loss: 0.7785 (0.7822)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [95]  [ 900/6250]  eta: 0:24:19  lr: 0.000001  grad: 0.2184 (0.4269)  loss: 0.7820 (0.7820)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [95]  [1000/6250]  eta: 0:23:49  lr: 0.000001  grad: 0.2067 (0.4308)  loss: 0.7789 (0.7821)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [95]  [1100/6250]  eta: 0:23:19  lr: 0.000001  grad: 0.2272 (0.4212)  loss: 0.7895 (0.7825)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [95]  [1200/6250]  eta: 0:22:50  lr: 0.000001  grad: 0.2483 (0.4179)  loss: 0.7812 (0.7825)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [95]  [1300/6250]  eta: 0:22:22  lr: 0.000001  grad: 0.2098 (0.4153)  loss: 0.7758 (0.7827)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [95]  [1400/6250]  eta: 0:21:53  lr: 0.000001  grad: 0.2248 (0.4065)  loss: 0.7774 (0.7826)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [95]  [1500/6250]  eta: 0:21:25  lr: 0.000001  grad: 0.2090 (0.4033)  loss: 0.7779 (0.7824)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [95]  [1600/6250]  eta: 0:20:56  lr: 0.000001  grad: 0.2247 (0.4034)  loss: 0.7796 (0.7823)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [95]  [1700/6250]  eta: 0:20:29  lr: 0.000001  grad: 0.2214 (0.3981)  loss: 0.7770 (0.7820)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [95]  [1800/6250]  eta: 0:20:01  lr: 0.000001  grad: 0.2220 (0.3912)  loss: 0.7820 (0.7818)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [95]  [1900/6250]  eta: 0:19:33  lr: 0.000001  grad: 0.2208 (0.3928)  loss: 0.7798 (0.7818)  time: 0.2660  data: 0.0002  max mem: 26157
Train: [95]  [2000/6250]  eta: 0:19:05  lr: 0.000001  grad: 0.2215 (0.3970)  loss: 0.7786 (0.7816)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [95]  [2100/6250]  eta: 0:18:38  lr: 0.000001  grad: 0.2121 (0.3947)  loss: 0.7847 (0.7815)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [95]  [2200/6250]  eta: 0:18:10  lr: 0.000001  grad: 0.2160 (0.4010)  loss: 0.7779 (0.7814)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [95]  [2300/6250]  eta: 0:17:43  lr: 0.000001  grad: 0.2300 (0.3990)  loss: 0.7833 (0.7812)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [95]  [2400/6250]  eta: 0:17:16  lr: 0.000001  grad: 0.2093 (0.3985)  loss: 0.7806 (0.7812)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [95]  [2500/6250]  eta: 0:16:49  lr: 0.000001  grad: 0.2228 (0.3978)  loss: 0.7859 (0.7813)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [95]  [2600/6250]  eta: 0:16:21  lr: 0.000001  grad: 0.2384 (0.4014)  loss: 0.7746 (0.7814)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [95]  [2700/6250]  eta: 0:15:55  lr: 0.000001  grad: 0.2192 (0.3979)  loss: 0.7819 (0.7814)  time: 0.2681  data: 0.0001  max mem: 26157
Train: [95]  [2800/6250]  eta: 0:15:28  lr: 0.000001  grad: 0.2108 (0.3967)  loss: 0.7859 (0.7815)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [95]  [2900/6250]  eta: 0:15:01  lr: 0.000001  grad: 0.2106 (0.3963)  loss: 0.7933 (0.7816)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [95]  [3000/6250]  eta: 0:14:34  lr: 0.000001  grad: 0.2115 (0.4006)  loss: 0.7716 (0.7816)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [95]  [3100/6250]  eta: 0:14:07  lr: 0.000001  grad: 0.2007 (0.4061)  loss: 0.7842 (0.7816)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [95]  [3200/6250]  eta: 0:13:40  lr: 0.000001  grad: 0.2109 (0.4106)  loss: 0.7877 (0.7817)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [95]  [3300/6250]  eta: 0:13:13  lr: 0.000001  grad: 0.2310 (0.4062)  loss: 0.7847 (0.7817)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [95]  [3400/6250]  eta: 0:12:46  lr: 0.000001  grad: 0.2292 (0.4061)  loss: 0.7726 (0.7817)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [95]  [3500/6250]  eta: 0:12:19  lr: 0.000001  grad: 0.2200 (0.4042)  loss: 0.7816 (0.7817)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [95]  [3600/6250]  eta: 0:11:52  lr: 0.000001  grad: 0.2294 (0.4029)  loss: 0.7677 (0.7817)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [95]  [3700/6250]  eta: 0:11:25  lr: 0.000001  grad: 0.2193 (0.4011)  loss: 0.7920 (0.7818)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [95]  [3800/6250]  eta: 0:10:58  lr: 0.000001  grad: 0.2178 (0.4009)  loss: 0.7842 (0.7818)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [95]  [3900/6250]  eta: 0:10:31  lr: 0.000001  grad: 0.2372 (0.3977)  loss: 0.7772 (0.7818)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [95]  [4000/6250]  eta: 0:10:04  lr: 0.000001  grad: 0.2357 (0.4010)  loss: 0.7757 (0.7818)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [95]  [4100/6250]  eta: 0:09:37  lr: 0.000001  grad: 0.2427 (0.4037)  loss: 0.7774 (0.7819)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [95]  [4200/6250]  eta: 0:09:10  lr: 0.000001  grad: 0.2322 (0.4039)  loss: 0.7737 (0.7819)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [95]  [4300/6250]  eta: 0:08:43  lr: 0.000001  grad: 0.2125 (0.4068)  loss: 0.7832 (0.7819)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [95]  [4400/6250]  eta: 0:08:16  lr: 0.000001  grad: 0.2047 (0.4069)  loss: 0.7880 (0.7820)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [95]  [4500/6250]  eta: 0:07:49  lr: 0.000001  grad: 0.2197 (0.4049)  loss: 0.7868 (0.7821)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [95]  [4600/6250]  eta: 0:07:22  lr: 0.000001  grad: 0.2145 (0.4036)  loss: 0.7873 (0.7822)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [95]  [4700/6250]  eta: 0:06:55  lr: 0.000001  grad: 0.2102 (0.4051)  loss: 0.7864 (0.7823)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [95]  [4800/6250]  eta: 0:06:29  lr: 0.000001  grad: 0.1985 (0.4025)  loss: 0.7908 (0.7824)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [95]  [4900/6250]  eta: 0:06:02  lr: 0.000001  grad: 0.2058 (0.4046)  loss: 0.7898 (0.7825)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [95]  [5000/6250]  eta: 0:05:35  lr: 0.000001  grad: 0.2150 (0.4084)  loss: 0.7825 (0.7826)  time: 0.2692  data: 0.0001  max mem: 26157
Train: [95]  [5100/6250]  eta: 0:05:08  lr: 0.000001  grad: 0.2333 (0.4064)  loss: 0.7897 (0.7827)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [95]  [5200/6250]  eta: 0:04:41  lr: 0.000001  grad: 0.2278 (0.4057)  loss: 0.7907 (0.7828)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [95]  [5300/6250]  eta: 0:04:14  lr: 0.000001  grad: 0.1950 (0.4045)  loss: 0.7949 (0.7828)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [95]  [5400/6250]  eta: 0:03:47  lr: 0.000001  grad: 0.2172 (0.4024)  loss: 0.7923 (0.7829)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [95]  [5500/6250]  eta: 0:03:21  lr: 0.000001  grad: 0.2081 (0.4041)  loss: 0.7885 (0.7831)  time: 0.2656  data: 0.0001  max mem: 26157
Train: [95]  [5600/6250]  eta: 0:02:54  lr: 0.000001  grad: 0.2132 (0.4024)  loss: 0.7920 (0.7832)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [95]  [5700/6250]  eta: 0:02:27  lr: 0.000001  grad: 0.2176 (0.4015)  loss: 0.7892 (0.7832)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [95]  [5800/6250]  eta: 0:02:00  lr: 0.000001  grad: 0.2150 (0.4004)  loss: 0.7876 (0.7833)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [95]  [5900/6250]  eta: 0:01:33  lr: 0.000001  grad: 0.2243 (0.3993)  loss: 0.7815 (0.7833)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [95]  [6000/6250]  eta: 0:01:07  lr: 0.000001  grad: 0.2380 (0.4046)  loss: 0.7846 (0.7833)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [95]  [6100/6250]  eta: 0:00:40  lr: 0.000001  grad: 0.2123 (0.4103)  loss: 0.7851 (0.7834)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [95]  [6200/6250]  eta: 0:00:13  lr: 0.000001  grad: 0.2339 (0.4129)  loss: 0.7811 (0.7833)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [95]  [6249/6250]  eta: 0:00:00  lr: 0.000001  grad: 0.2143 (0.4138)  loss: 0.7844 (0.7833)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [95] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000001  grad: 0.2143 (0.4138)  loss: 0.7844 (0.7833)
Eval (hcp-train-subset): [95]  [ 0/62]  eta: 0:05:05  loss: 0.7847 (0.7847)  time: 4.9198  data: 4.8369  max mem: 26157
Eval (hcp-train-subset): [95]  [61/62]  eta: 0:00:00  loss: 0.7800 (0.7849)  time: 0.1027  data: 0.0202  max mem: 26157
Eval (hcp-train-subset): [95] Total time: 0:00:11 (0.1883 s / it)
Averaged stats (hcp-train-subset): loss: 0.7800 (0.7849)
Making plots (hcp-train-subset): example=16
Eval (hcp-val): [95]  [ 0/62]  eta: 0:03:50  loss: 0.8171 (0.8171)  time: 3.7161  data: 3.6109  max mem: 26157
Eval (hcp-val): [95]  [61/62]  eta: 0:00:00  loss: 0.8218 (0.8218)  time: 0.1082  data: 0.0257  max mem: 26157
Eval (hcp-val): [95] Total time: 0:00:11 (0.1854 s / it)
Averaged stats (hcp-val): loss: 0.8218 (0.8218)
Making plots (hcp-val): example=40
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [96]  [   0/6250]  eta: 8:24:52  lr: 0.000001  grad: 0.3420 (0.3420)  loss: 0.7692 (0.7692)  time: 4.8467  data: 4.5758  max mem: 26157
Train: [96]  [ 100/6250]  eta: 0:32:02  lr: 0.000001  grad: 0.2700 (0.6761)  loss: 0.7954 (0.7945)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [96]  [ 200/6250]  eta: 0:29:12  lr: 0.000001  grad: 0.2459 (0.4901)  loss: 0.7910 (0.7959)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [96]  [ 300/6250]  eta: 0:27:58  lr: 0.000001  grad: 0.2594 (0.4154)  loss: 0.8029 (0.7970)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [96]  [ 400/6250]  eta: 0:27:09  lr: 0.000001  grad: 0.2429 (0.3975)  loss: 0.7974 (0.7971)  time: 0.2675  data: 0.0002  max mem: 26157
Train: [96]  [ 500/6250]  eta: 0:26:28  lr: 0.000001  grad: 0.2304 (0.3853)  loss: 0.7925 (0.7964)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [96]  [ 600/6250]  eta: 0:25:52  lr: 0.000001  grad: 0.2320 (0.3827)  loss: 0.7797 (0.7944)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [96]  [ 700/6250]  eta: 0:25:18  lr: 0.000001  grad: 0.2234 (0.3700)  loss: 0.7873 (0.7931)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [96]  [ 800/6250]  eta: 0:24:47  lr: 0.000001  grad: 0.2302 (0.3754)  loss: 0.7791 (0.7920)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [96]  [ 900/6250]  eta: 0:24:16  lr: 0.000001  grad: 0.2381 (0.3979)  loss: 0.7796 (0.7909)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [96]  [1000/6250]  eta: 0:23:46  lr: 0.000001  grad: 0.2230 (0.3882)  loss: 0.7874 (0.7902)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [96]  [1100/6250]  eta: 0:23:17  lr: 0.000000  grad: 0.2290 (0.4313)  loss: 0.7852 (0.7896)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [96]  [1200/6250]  eta: 0:22:48  lr: 0.000000  grad: 0.2231 (0.4237)  loss: 0.7792 (0.7891)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [96]  [1300/6250]  eta: 0:22:19  lr: 0.000000  grad: 0.2409 (0.4201)  loss: 0.7810 (0.7886)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [96]  [1400/6250]  eta: 0:21:51  lr: 0.000000  grad: 0.2366 (0.4238)  loss: 0.7854 (0.7883)  time: 0.2680  data: 0.0002  max mem: 26157
Train: [96]  [1500/6250]  eta: 0:21:23  lr: 0.000000  grad: 0.2287 (0.4191)  loss: 0.7727 (0.7878)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [96]  [1600/6250]  eta: 0:20:55  lr: 0.000000  grad: 0.2423 (0.4148)  loss: 0.7842 (0.7875)  time: 0.2663  data: 0.0002  max mem: 26157
Train: [96]  [1700/6250]  eta: 0:20:27  lr: 0.000000  grad: 0.2420 (0.4110)  loss: 0.7784 (0.7870)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [96]  [1800/6250]  eta: 0:19:59  lr: 0.000000  grad: 0.2384 (0.4107)  loss: 0.7815 (0.7866)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [96]  [1900/6250]  eta: 0:19:32  lr: 0.000000  grad: 0.2270 (0.4154)  loss: 0.7744 (0.7861)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [96]  [2000/6250]  eta: 0:19:04  lr: 0.000000  grad: 0.2147 (0.4074)  loss: 0.7779 (0.7858)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [96]  [2100/6250]  eta: 0:18:37  lr: 0.000000  grad: 0.2276 (0.4206)  loss: 0.7805 (0.7854)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [96]  [2200/6250]  eta: 0:18:10  lr: 0.000000  grad: 0.2080 (0.4150)  loss: 0.7852 (0.7852)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [96]  [2300/6250]  eta: 0:17:42  lr: 0.000000  grad: 0.2311 (0.4116)  loss: 0.7704 (0.7848)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [96]  [2400/6250]  eta: 0:17:15  lr: 0.000000  grad: 0.2313 (0.4144)  loss: 0.7785 (0.7847)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [96]  [2500/6250]  eta: 0:16:48  lr: 0.000000  grad: 0.2221 (0.4172)  loss: 0.7820 (0.7846)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [96]  [2600/6250]  eta: 0:16:21  lr: 0.000000  grad: 0.2233 (0.4148)  loss: 0.7851 (0.7846)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [96]  [2700/6250]  eta: 0:15:53  lr: 0.000000  grad: 0.2351 (0.4119)  loss: 0.7767 (0.7845)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [96]  [2800/6250]  eta: 0:15:26  lr: 0.000000  grad: 0.2364 (0.4107)  loss: 0.7780 (0.7844)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [96]  [2900/6250]  eta: 0:14:59  lr: 0.000000  grad: 0.2207 (0.4090)  loss: 0.7797 (0.7842)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [96]  [3000/6250]  eta: 0:14:32  lr: 0.000000  grad: 0.2234 (0.4052)  loss: 0.7789 (0.7840)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [96]  [3100/6250]  eta: 0:14:05  lr: 0.000000  grad: 0.2267 (0.4068)  loss: 0.7723 (0.7840)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [96]  [3200/6250]  eta: 0:13:38  lr: 0.000000  grad: 0.2155 (0.4126)  loss: 0.7783 (0.7838)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [96]  [3300/6250]  eta: 0:13:11  lr: 0.000000  grad: 0.2101 (0.4093)  loss: 0.7766 (0.7837)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [96]  [3400/6250]  eta: 0:12:44  lr: 0.000000  grad: 0.2156 (0.4062)  loss: 0.7830 (0.7836)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [96]  [3500/6250]  eta: 0:12:17  lr: 0.000000  grad: 0.2219 (0.4132)  loss: 0.7829 (0.7835)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [96]  [3600/6250]  eta: 0:11:50  lr: 0.000000  grad: 0.2226 (0.4105)  loss: 0.7809 (0.7834)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [96]  [3700/6250]  eta: 0:11:24  lr: 0.000000  grad: 0.2210 (0.4123)  loss: 0.7783 (0.7833)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [96]  [3800/6250]  eta: 0:10:57  lr: 0.000000  grad: 0.2054 (0.4096)  loss: 0.7809 (0.7832)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [96]  [3900/6250]  eta: 0:10:30  lr: 0.000000  grad: 0.2191 (0.4054)  loss: 0.7673 (0.7831)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [96]  [4000/6250]  eta: 0:10:03  lr: 0.000000  grad: 0.2309 (0.4046)  loss: 0.7816 (0.7830)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [96]  [4100/6250]  eta: 0:09:36  lr: 0.000000  grad: 0.2341 (0.4008)  loss: 0.7808 (0.7829)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [96]  [4200/6250]  eta: 0:09:09  lr: 0.000000  grad: 0.2170 (0.3988)  loss: 0.7835 (0.7828)  time: 0.2659  data: 0.0001  max mem: 26157
Train: [96]  [4300/6250]  eta: 0:08:42  lr: 0.000000  grad: 0.2273 (0.3970)  loss: 0.7737 (0.7827)  time: 0.2693  data: 0.0002  max mem: 26157
Train: [96]  [4400/6250]  eta: 0:08:15  lr: 0.000000  grad: 0.2269 (0.4009)  loss: 0.7812 (0.7826)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [96]  [4500/6250]  eta: 0:07:49  lr: 0.000000  grad: 0.2445 (0.4026)  loss: 0.7739 (0.7825)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [96]  [4600/6250]  eta: 0:07:22  lr: 0.000000  grad: 0.2410 (0.4039)  loss: 0.7807 (0.7824)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [96]  [4700/6250]  eta: 0:06:55  lr: 0.000000  grad: 0.2584 (0.4026)  loss: 0.7811 (0.7824)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [96]  [4800/6250]  eta: 0:06:28  lr: 0.000000  grad: 0.2243 (0.4007)  loss: 0.7745 (0.7823)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [96]  [4900/6250]  eta: 0:06:01  lr: 0.000000  grad: 0.2505 (0.4044)  loss: 0.7730 (0.7822)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [96]  [5000/6250]  eta: 0:05:34  lr: 0.000000  grad: 0.2566 (0.4062)  loss: 0.7830 (0.7822)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [96]  [5100/6250]  eta: 0:05:08  lr: 0.000000  grad: 0.2395 (0.4051)  loss: 0.7789 (0.7821)  time: 0.2684  data: 0.0002  max mem: 26157
Train: [96]  [5200/6250]  eta: 0:04:41  lr: 0.000000  grad: 0.2330 (0.4032)  loss: 0.7898 (0.7821)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [96]  [5300/6250]  eta: 0:04:14  lr: 0.000000  grad: 0.2346 (0.4048)  loss: 0.7834 (0.7821)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [96]  [5400/6250]  eta: 0:03:47  lr: 0.000000  grad: 0.2304 (0.4048)  loss: 0.7753 (0.7820)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [96]  [5500/6250]  eta: 0:03:20  lr: 0.000000  grad: 0.2307 (0.4046)  loss: 0.7753 (0.7820)  time: 0.2658  data: 0.0001  max mem: 26157
Train: [96]  [5600/6250]  eta: 0:02:54  lr: 0.000000  grad: 0.2271 (0.4042)  loss: 0.7826 (0.7820)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [96]  [5700/6250]  eta: 0:02:27  lr: 0.000000  grad: 0.2240 (0.4030)  loss: 0.7798 (0.7819)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [96]  [5800/6250]  eta: 0:02:00  lr: 0.000000  grad: 0.2219 (0.4043)  loss: 0.7830 (0.7819)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [96]  [5900/6250]  eta: 0:01:33  lr: 0.000000  grad: 0.2319 (0.4052)  loss: 0.7797 (0.7819)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [96]  [6000/6250]  eta: 0:01:06  lr: 0.000000  grad: 0.2209 (0.4056)  loss: 0.7800 (0.7819)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [96]  [6100/6250]  eta: 0:00:40  lr: 0.000000  grad: 0.2474 (0.4056)  loss: 0.7865 (0.7819)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [96]  [6200/6250]  eta: 0:00:13  lr: 0.000000  grad: 0.2462 (0.4105)  loss: 0.7828 (0.7819)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [96]  [6249/6250]  eta: 0:00:00  lr: 0.000000  grad: 0.2159 (0.4125)  loss: 0.7834 (0.7819)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [96] Total time: 0:27:57 (0.2683 s / it)
Averaged stats: lr: 0.000000  grad: 0.2159 (0.4125)  loss: 0.7834 (0.7819)
Eval (hcp-train-subset): [96]  [ 0/62]  eta: 0:04:02  loss: 0.7845 (0.7845)  time: 3.9101  data: 3.8275  max mem: 26157
Eval (hcp-train-subset): [96]  [61/62]  eta: 0:00:00  loss: 0.7801 (0.7842)  time: 0.0822  data: 0.0001  max mem: 26157
Eval (hcp-train-subset): [96] Total time: 0:00:10 (0.1649 s / it)
Averaged stats (hcp-train-subset): loss: 0.7801 (0.7842)
Making plots (hcp-train-subset): example=40
Eval (hcp-val): [96]  [ 0/62]  eta: 0:04:36  loss: 0.8175 (0.8175)  time: 4.4570  data: 4.3737  max mem: 26157
Eval (hcp-val): [96]  [61/62]  eta: 0:00:00  loss: 0.8231 (0.8224)  time: 0.0841  data: 0.0001  max mem: 26157
Eval (hcp-val): [96] Total time: 0:00:10 (0.1706 s / it)
Averaged stats (hcp-val): loss: 0.8231 (0.8224)
Making plots (hcp-val): example=55
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [97]  [   0/6250]  eta: 6:25:49  lr: 0.000000  grad: 0.2990 (0.2990)  loss: 0.8480 (0.8480)  time: 3.7039  data: 3.3666  max mem: 26157
Train: [97]  [ 100/6250]  eta: 0:32:28  lr: 0.000000  grad: 0.2720 (0.4532)  loss: 0.7932 (0.7922)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [97]  [ 200/6250]  eta: 0:29:25  lr: 0.000000  grad: 0.2916 (0.4127)  loss: 0.7965 (0.7910)  time: 0.2666  data: 0.0002  max mem: 26157
Train: [97]  [ 300/6250]  eta: 0:28:08  lr: 0.000000  grad: 0.2338 (0.3919)  loss: 0.8031 (0.7925)  time: 0.2683  data: 0.0001  max mem: 26157
Train: [97]  [ 400/6250]  eta: 0:27:16  lr: 0.000000  grad: 0.2335 (0.3670)  loss: 0.7918 (0.7942)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [97]  [ 500/6250]  eta: 0:26:34  lr: 0.000000  grad: 0.2291 (0.3701)  loss: 0.7883 (0.7937)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [97]  [ 600/6250]  eta: 0:25:57  lr: 0.000000  grad: 0.2308 (0.3727)  loss: 0.7911 (0.7931)  time: 0.2680  data: 0.0001  max mem: 26157
Train: [97]  [ 700/6250]  eta: 0:25:23  lr: 0.000000  grad: 0.2290 (0.3644)  loss: 0.7824 (0.7923)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [97]  [ 800/6250]  eta: 0:24:50  lr: 0.000000  grad: 0.2218 (0.3517)  loss: 0.7830 (0.7918)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [97]  [ 900/6250]  eta: 0:24:19  lr: 0.000000  grad: 0.2256 (0.3438)  loss: 0.7916 (0.7913)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [97]  [1000/6250]  eta: 0:23:49  lr: 0.000000  grad: 0.2336 (0.3419)  loss: 0.7768 (0.7910)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [97]  [1100/6250]  eta: 0:23:19  lr: 0.000000  grad: 0.2216 (0.3499)  loss: 0.7855 (0.7906)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [97]  [1200/6250]  eta: 0:22:50  lr: 0.000000  grad: 0.2189 (0.3583)  loss: 0.7826 (0.7903)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [97]  [1300/6250]  eta: 0:22:22  lr: 0.000000  grad: 0.2131 (0.3533)  loss: 0.7919 (0.7902)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [97]  [1400/6250]  eta: 0:21:53  lr: 0.000000  grad: 0.2068 (0.3490)  loss: 0.7879 (0.7900)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [97]  [1500/6250]  eta: 0:21:25  lr: 0.000000  grad: 0.2230 (0.3507)  loss: 0.7792 (0.7894)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [97]  [1600/6250]  eta: 0:20:57  lr: 0.000000  grad: 0.2406 (0.3499)  loss: 0.7890 (0.7892)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [97]  [1700/6250]  eta: 0:20:29  lr: 0.000000  grad: 0.2213 (0.3469)  loss: 0.7785 (0.7889)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [97]  [1800/6250]  eta: 0:20:01  lr: 0.000000  grad: 0.2229 (0.3487)  loss: 0.7856 (0.7887)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [97]  [1900/6250]  eta: 0:19:33  lr: 0.000000  grad: 0.2373 (0.3520)  loss: 0.7842 (0.7887)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [97]  [2000/6250]  eta: 0:19:06  lr: 0.000000  grad: 0.2281 (0.3616)  loss: 0.7853 (0.7885)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [97]  [2100/6250]  eta: 0:18:38  lr: 0.000000  grad: 0.2173 (0.3608)  loss: 0.7814 (0.7885)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [97]  [2200/6250]  eta: 0:18:11  lr: 0.000000  grad: 0.2168 (0.3667)  loss: 0.7966 (0.7886)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [97]  [2300/6250]  eta: 0:17:43  lr: 0.000000  grad: 0.2324 (0.3643)  loss: 0.7896 (0.7887)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [97]  [2400/6250]  eta: 0:17:16  lr: 0.000000  grad: 0.2282 (0.3644)  loss: 0.7862 (0.7886)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [97]  [2500/6250]  eta: 0:16:49  lr: 0.000000  grad: 0.2230 (0.3668)  loss: 0.7859 (0.7886)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [97]  [2600/6250]  eta: 0:16:22  lr: 0.000000  grad: 0.2033 (0.3649)  loss: 0.7937 (0.7884)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [97]  [2700/6250]  eta: 0:15:54  lr: 0.000000  grad: 0.2255 (0.3628)  loss: 0.7828 (0.7883)  time: 0.2657  data: 0.0001  max mem: 26157
Train: [97]  [2800/6250]  eta: 0:15:27  lr: 0.000000  grad: 0.2263 (0.3600)  loss: 0.7806 (0.7881)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [97]  [2900/6250]  eta: 0:15:00  lr: 0.000000  grad: 0.2264 (0.3589)  loss: 0.7787 (0.7879)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [97]  [3000/6250]  eta: 0:14:33  lr: 0.000000  grad: 0.2304 (0.3585)  loss: 0.7862 (0.7879)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [97]  [3100/6250]  eta: 0:14:06  lr: 0.000000  grad: 0.2160 (0.3575)  loss: 0.7851 (0.7877)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [97]  [3200/6250]  eta: 0:13:39  lr: 0.000000  grad: 0.2192 (0.3620)  loss: 0.7845 (0.7875)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [97]  [3300/6250]  eta: 0:13:12  lr: 0.000000  grad: 0.2409 (0.3643)  loss: 0.7887 (0.7875)  time: 0.2678  data: 0.0002  max mem: 26157
Train: [97]  [3400/6250]  eta: 0:12:45  lr: 0.000000  grad: 0.2503 (0.3672)  loss: 0.7821 (0.7874)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [97]  [3500/6250]  eta: 0:12:18  lr: 0.000000  grad: 0.2359 (0.3717)  loss: 0.7812 (0.7873)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [97]  [3600/6250]  eta: 0:11:51  lr: 0.000000  grad: 0.2396 (0.3711)  loss: 0.7827 (0.7873)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [97]  [3700/6250]  eta: 0:11:24  lr: 0.000000  grad: 0.2009 (0.3746)  loss: 0.7816 (0.7872)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [97]  [3800/6250]  eta: 0:10:57  lr: 0.000000  grad: 0.2376 (0.3720)  loss: 0.7867 (0.7872)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [97]  [3900/6250]  eta: 0:10:30  lr: 0.000000  grad: 0.2327 (0.3734)  loss: 0.7825 (0.7872)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [97]  [4000/6250]  eta: 0:10:03  lr: 0.000000  grad: 0.2175 (0.3742)  loss: 0.7818 (0.7872)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [97]  [4100/6250]  eta: 0:09:36  lr: 0.000000  grad: 0.2113 (0.3753)  loss: 0.7819 (0.7872)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [97]  [4200/6250]  eta: 0:09:09  lr: 0.000000  grad: 0.2229 (0.3740)  loss: 0.7791 (0.7871)  time: 0.2702  data: 0.0002  max mem: 26157
Train: [97]  [4300/6250]  eta: 0:08:42  lr: 0.000000  grad: 0.2375 (0.3750)  loss: 0.7815 (0.7870)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [97]  [4400/6250]  eta: 0:08:16  lr: 0.000000  grad: 0.2170 (0.3742)  loss: 0.7859 (0.7869)  time: 0.2711  data: 0.0002  max mem: 26157
Train: [97]  [4500/6250]  eta: 0:07:49  lr: 0.000000  grad: 0.2183 (0.3824)  loss: 0.7812 (0.7868)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [97]  [4600/6250]  eta: 0:07:22  lr: 0.000000  grad: 0.2111 (0.3796)  loss: 0.7899 (0.7868)  time: 0.2669  data: 0.0002  max mem: 26157
Train: [97]  [4700/6250]  eta: 0:06:55  lr: 0.000000  grad: 0.2295 (0.3796)  loss: 0.7888 (0.7867)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [97]  [4800/6250]  eta: 0:06:28  lr: 0.000000  grad: 0.2221 (0.3862)  loss: 0.7867 (0.7866)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [97]  [4900/6250]  eta: 0:06:01  lr: 0.000000  grad: 0.2159 (0.3848)  loss: 0.7887 (0.7865)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [97]  [5000/6250]  eta: 0:05:35  lr: 0.000000  grad: 0.2216 (0.3909)  loss: 0.7827 (0.7864)  time: 0.2673  data: 0.0001  max mem: 26157
Train: [97]  [5100/6250]  eta: 0:05:08  lr: 0.000000  grad: 0.2327 (0.3908)  loss: 0.7755 (0.7864)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [97]  [5200/6250]  eta: 0:04:41  lr: 0.000000  grad: 0.2283 (0.3912)  loss: 0.7799 (0.7863)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [97]  [5300/6250]  eta: 0:04:14  lr: 0.000000  grad: 0.2425 (0.3921)  loss: 0.7787 (0.7863)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [97]  [5400/6250]  eta: 0:03:47  lr: 0.000000  grad: 0.2111 (0.3902)  loss: 0.7931 (0.7863)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [97]  [5500/6250]  eta: 0:03:21  lr: 0.000000  grad: 0.2229 (0.3907)  loss: 0.7916 (0.7863)  time: 0.2688  data: 0.0002  max mem: 26157
Train: [97]  [5600/6250]  eta: 0:02:54  lr: 0.000000  grad: 0.2357 (0.3906)  loss: 0.7936 (0.7863)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [97]  [5700/6250]  eta: 0:02:27  lr: 0.000000  grad: 0.2242 (0.3933)  loss: 0.7878 (0.7863)  time: 0.2663  data: 0.0002  max mem: 26157
Train: [97]  [5800/6250]  eta: 0:02:00  lr: 0.000000  grad: 0.2325 (0.3917)  loss: 0.7879 (0.7864)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [97]  [5900/6250]  eta: 0:01:33  lr: 0.000000  grad: 0.2411 (0.3955)  loss: 0.7887 (0.7864)  time: 0.2676  data: 0.0001  max mem: 26157
Train: [97]  [6000/6250]  eta: 0:01:06  lr: 0.000000  grad: 0.2288 (0.3945)  loss: 0.7916 (0.7864)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [97]  [6100/6250]  eta: 0:00:40  lr: 0.000000  grad: 0.2364 (0.3931)  loss: 0.7906 (0.7864)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [97]  [6200/6250]  eta: 0:00:13  lr: 0.000000  grad: 0.2269 (0.3909)  loss: 0.7878 (0.7864)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [97]  [6249/6250]  eta: 0:00:00  lr: 0.000000  grad: 0.2358 (0.3898)  loss: 0.7849 (0.7864)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [97] Total time: 0:27:57 (0.2684 s / it)
Averaged stats: lr: 0.000000  grad: 0.2358 (0.3898)  loss: 0.7849 (0.7864)
Eval (hcp-train-subset): [97]  [ 0/62]  eta: 0:04:20  loss: 0.7849 (0.7849)  time: 4.2078  data: 4.1247  max mem: 26157
Eval (hcp-train-subset): [97]  [61/62]  eta: 0:00:00  loss: 0.7781 (0.7838)  time: 0.0947  data: 0.0127  max mem: 26157
Eval (hcp-train-subset): [97] Total time: 0:00:10 (0.1689 s / it)
Averaged stats (hcp-train-subset): loss: 0.7781 (0.7838)
Making plots (hcp-train-subset): example=17
Eval (hcp-val): [97]  [ 0/62]  eta: 0:04:33  loss: 0.8193 (0.8193)  time: 4.4086  data: 4.3257  max mem: 26157
Eval (hcp-val): [97]  [61/62]  eta: 0:00:00  loss: 0.8216 (0.8224)  time: 0.0923  data: 0.0084  max mem: 26157
Eval (hcp-val): [97] Total time: 0:00:10 (0.1679 s / it)
Averaged stats (hcp-val): loss: 0.8216 (0.8224)
Making plots (hcp-val): example=50
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [98]  [   0/6250]  eta: 8:38:52  lr: 0.000000  grad: 0.5456 (0.5456)  loss: 0.8337 (0.8337)  time: 4.9811  data: 4.6967  max mem: 26157
Train: [98]  [ 100/6250]  eta: 0:33:05  lr: 0.000000  grad: 0.3186 (0.6388)  loss: 0.7771 (0.7832)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [98]  [ 200/6250]  eta: 0:29:45  lr: 0.000000  grad: 0.2837 (0.4885)  loss: 0.7871 (0.7842)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [98]  [ 300/6250]  eta: 0:28:19  lr: 0.000000  grad: 0.2450 (0.4376)  loss: 0.7869 (0.7853)  time: 0.2660  data: 0.0001  max mem: 26157
Train: [98]  [ 400/6250]  eta: 0:27:23  lr: 0.000000  grad: 0.2126 (0.3964)  loss: 0.7997 (0.7881)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [98]  [ 500/6250]  eta: 0:26:39  lr: 0.000000  grad: 0.2220 (0.4066)  loss: 0.7960 (0.7879)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [98]  [ 600/6250]  eta: 0:26:01  lr: 0.000000  grad: 0.2432 (0.4181)  loss: 0.7848 (0.7872)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [98]  [ 700/6250]  eta: 0:25:26  lr: 0.000000  grad: 0.2535 (0.4379)  loss: 0.7897 (0.7871)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [98]  [ 800/6250]  eta: 0:24:53  lr: 0.000000  grad: 0.2411 (0.4345)  loss: 0.7854 (0.7868)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [98]  [ 900/6250]  eta: 0:24:22  lr: 0.000000  grad: 0.2259 (0.4232)  loss: 0.8004 (0.7873)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [98]  [1000/6250]  eta: 0:23:51  lr: 0.000000  grad: 0.2292 (0.4092)  loss: 0.7856 (0.7872)  time: 0.2686  data: 0.0002  max mem: 26157
Train: [98]  [1100/6250]  eta: 0:23:21  lr: 0.000000  grad: 0.2311 (0.4063)  loss: 0.7757 (0.7869)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [1200/6250]  eta: 0:22:52  lr: 0.000000  grad: 0.2301 (0.4034)  loss: 0.7997 (0.7870)  time: 0.2682  data: 0.0002  max mem: 26157
Train: [98]  [1300/6250]  eta: 0:22:23  lr: 0.000000  grad: 0.2396 (0.3953)  loss: 0.7925 (0.7871)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [98]  [1400/6250]  eta: 0:21:54  lr: 0.000000  grad: 0.2395 (0.3994)  loss: 0.7850 (0.7871)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [98]  [1500/6250]  eta: 0:21:26  lr: 0.000000  grad: 0.2226 (0.3924)  loss: 0.7844 (0.7873)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [1600/6250]  eta: 0:20:58  lr: 0.000000  grad: 0.2422 (0.3957)  loss: 0.7878 (0.7875)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [98]  [1700/6250]  eta: 0:20:30  lr: 0.000000  grad: 0.2250 (0.3977)  loss: 0.7906 (0.7875)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [98]  [1800/6250]  eta: 0:20:02  lr: 0.000000  grad: 0.2191 (0.3957)  loss: 0.7922 (0.7876)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [98]  [1900/6250]  eta: 0:19:34  lr: 0.000000  grad: 0.2093 (0.3974)  loss: 0.7915 (0.7877)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [98]  [2000/6250]  eta: 0:19:07  lr: 0.000000  grad: 0.2069 (0.3951)  loss: 0.7958 (0.7879)  time: 0.2674  data: 0.0002  max mem: 26157
Train: [98]  [2100/6250]  eta: 0:18:39  lr: 0.000000  grad: 0.2033 (0.3888)  loss: 0.7874 (0.7880)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [2200/6250]  eta: 0:18:12  lr: 0.000000  grad: 0.2244 (0.3947)  loss: 0.7850 (0.7880)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [98]  [2300/6250]  eta: 0:17:44  lr: 0.000000  grad: 0.2166 (0.3934)  loss: 0.7861 (0.7881)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [98]  [2400/6250]  eta: 0:17:17  lr: 0.000000  grad: 0.2080 (0.3934)  loss: 0.7887 (0.7882)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [98]  [2500/6250]  eta: 0:16:50  lr: 0.000000  grad: 0.2189 (0.4045)  loss: 0.7876 (0.7881)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [98]  [2600/6250]  eta: 0:16:22  lr: 0.000000  grad: 0.2177 (0.4032)  loss: 0.7929 (0.7881)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [98]  [2700/6250]  eta: 0:15:55  lr: 0.000000  grad: 0.2304 (0.4049)  loss: 0.7860 (0.7880)  time: 0.2661  data: 0.0001  max mem: 26157
Train: [98]  [2800/6250]  eta: 0:15:28  lr: 0.000000  grad: 0.2190 (0.4038)  loss: 0.7853 (0.7879)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [98]  [2900/6250]  eta: 0:15:01  lr: 0.000000  grad: 0.2260 (0.4021)  loss: 0.7848 (0.7879)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [98]  [3000/6250]  eta: 0:14:34  lr: 0.000000  grad: 0.2391 (0.4018)  loss: 0.7812 (0.7878)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [98]  [3100/6250]  eta: 0:14:07  lr: 0.000000  grad: 0.2189 (0.4046)  loss: 0.7828 (0.7876)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [98]  [3200/6250]  eta: 0:13:40  lr: 0.000000  grad: 0.2347 (0.4045)  loss: 0.7834 (0.7874)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [98]  [3300/6250]  eta: 0:13:13  lr: 0.000000  grad: 0.2180 (0.4021)  loss: 0.7847 (0.7874)  time: 0.2689  data: 0.0001  max mem: 26157
Train: [98]  [3400/6250]  eta: 0:12:46  lr: 0.000000  grad: 0.2136 (0.4006)  loss: 0.7870 (0.7874)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [98]  [3500/6250]  eta: 0:12:19  lr: 0.000000  grad: 0.2401 (0.3987)  loss: 0.7864 (0.7873)  time: 0.2679  data: 0.0001  max mem: 26157
Train: [98]  [3600/6250]  eta: 0:11:52  lr: 0.000000  grad: 0.2222 (0.4021)  loss: 0.7789 (0.7872)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [98]  [3700/6250]  eta: 0:11:25  lr: 0.000000  grad: 0.2110 (0.4027)  loss: 0.7833 (0.7872)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [98]  [3800/6250]  eta: 0:10:58  lr: 0.000000  grad: 0.2167 (0.4061)  loss: 0.7912 (0.7871)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [3900/6250]  eta: 0:10:31  lr: 0.000000  grad: 0.2170 (0.4104)  loss: 0.7895 (0.7871)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [98]  [4000/6250]  eta: 0:10:04  lr: 0.000000  grad: 0.2247 (0.4083)  loss: 0.7845 (0.7871)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [98]  [4100/6250]  eta: 0:09:37  lr: 0.000000  grad: 0.2086 (0.4093)  loss: 0.7853 (0.7871)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [98]  [4200/6250]  eta: 0:09:10  lr: 0.000000  grad: 0.2215 (0.4074)  loss: 0.7802 (0.7870)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [98]  [4300/6250]  eta: 0:08:43  lr: 0.000000  grad: 0.2219 (0.4087)  loss: 0.7853 (0.7871)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [98]  [4400/6250]  eta: 0:08:16  lr: 0.000000  grad: 0.2350 (0.4049)  loss: 0.7896 (0.7871)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [98]  [4500/6250]  eta: 0:07:49  lr: 0.000000  grad: 0.2351 (0.4074)  loss: 0.7937 (0.7870)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [98]  [4600/6250]  eta: 0:07:22  lr: 0.000000  grad: 0.2236 (0.4044)  loss: 0.7847 (0.7870)  time: 0.2665  data: 0.0002  max mem: 26157
Train: [98]  [4700/6250]  eta: 0:06:55  lr: 0.000000  grad: 0.2239 (0.4027)  loss: 0.7843 (0.7869)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [4800/6250]  eta: 0:06:28  lr: 0.000000  grad: 0.2362 (0.4034)  loss: 0.7809 (0.7869)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [98]  [4900/6250]  eta: 0:06:02  lr: 0.000000  grad: 0.2228 (0.4019)  loss: 0.7862 (0.7869)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [98]  [5000/6250]  eta: 0:05:35  lr: 0.000000  grad: 0.2208 (0.4003)  loss: 0.7816 (0.7869)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [98]  [5100/6250]  eta: 0:05:08  lr: 0.000000  grad: 0.2167 (0.3977)  loss: 0.7842 (0.7869)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [98]  [5200/6250]  eta: 0:04:41  lr: 0.000000  grad: 0.2411 (0.3979)  loss: 0.7823 (0.7868)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [98]  [5300/6250]  eta: 0:04:14  lr: 0.000000  grad: 0.2016 (0.3971)  loss: 0.7883 (0.7868)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [98]  [5400/6250]  eta: 0:03:47  lr: 0.000000  grad: 0.2283 (0.3968)  loss: 0.7881 (0.7868)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [5500/6250]  eta: 0:03:21  lr: 0.000000  grad: 0.2266 (0.3956)  loss: 0.7907 (0.7868)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [98]  [5600/6250]  eta: 0:02:54  lr: 0.000000  grad: 0.2393 (0.3933)  loss: 0.7866 (0.7868)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [98]  [5700/6250]  eta: 0:02:27  lr: 0.000000  grad: 0.2249 (0.3954)  loss: 0.7851 (0.7868)  time: 0.2668  data: 0.0002  max mem: 26157
Train: [98]  [5800/6250]  eta: 0:02:00  lr: 0.000000  grad: 0.2351 (0.3987)  loss: 0.7828 (0.7868)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [5900/6250]  eta: 0:01:33  lr: 0.000000  grad: 0.2383 (0.3989)  loss: 0.7866 (0.7868)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [98]  [6000/6250]  eta: 0:01:07  lr: 0.000000  grad: 0.2187 (0.3977)  loss: 0.7826 (0.7868)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [98]  [6100/6250]  eta: 0:00:40  lr: 0.000000  grad: 0.2140 (0.3965)  loss: 0.7930 (0.7868)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [6200/6250]  eta: 0:00:13  lr: 0.000000  grad: 0.2126 (0.3986)  loss: 0.7946 (0.7869)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [98]  [6249/6250]  eta: 0:00:00  lr: 0.000000  grad: 0.2208 (0.3981)  loss: 0.7855 (0.7869)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [98] Total time: 0:27:58 (0.2685 s / it)
Averaged stats: lr: 0.000000  grad: 0.2208 (0.3981)  loss: 0.7855 (0.7869)
Eval (hcp-train-subset): [98]  [ 0/62]  eta: 0:04:31  loss: 0.7857 (0.7857)  time: 4.3772  data: 4.2943  max mem: 26157
Eval (hcp-train-subset): [98]  [61/62]  eta: 0:00:00  loss: 0.7796 (0.7839)  time: 0.0876  data: 0.0053  max mem: 26157
Eval (hcp-train-subset): [98] Total time: 0:00:10 (0.1685 s / it)
Averaged stats (hcp-train-subset): loss: 0.7796 (0.7839)
Making plots (hcp-train-subset): example=3
Eval (hcp-val): [98]  [ 0/62]  eta: 0:03:27  loss: 0.8152 (0.8152)  time: 3.3483  data: 3.2522  max mem: 26157
Eval (hcp-val): [98]  [61/62]  eta: 0:00:00  loss: 0.8208 (0.8218)  time: 0.0886  data: 0.0068  max mem: 26157
Eval (hcp-val): [98] Total time: 0:00:10 (0.1684 s / it)
Averaged stats (hcp-val): loss: 0.8208 (0.8218)
Making plots (hcp-val): example=32
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
Train: [99]  [   0/6250]  eta: 6:24:00  lr: 0.000000  grad: 0.2214 (0.2214)  loss: 0.8396 (0.8396)  time: 3.6865  data: 3.3715  max mem: 26157
Train: [99]  [ 100/6250]  eta: 0:31:11  lr: 0.000000  grad: 0.2967 (0.3555)  loss: 0.7971 (0.8034)  time: 0.2684  data: 0.0001  max mem: 26157
Train: [99]  [ 200/6250]  eta: 0:28:48  lr: 0.000000  grad: 0.2607 (0.3916)  loss: 0.8062 (0.8031)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [99]  [ 300/6250]  eta: 0:27:44  lr: 0.000000  grad: 0.2677 (0.3983)  loss: 0.7878 (0.7996)  time: 0.2671  data: 0.0002  max mem: 26157
Train: [99]  [ 400/6250]  eta: 0:26:57  lr: 0.000000  grad: 0.2698 (0.3808)  loss: 0.7833 (0.7962)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [99]  [ 500/6250]  eta: 0:26:18  lr: 0.000000  grad: 0.2739 (0.3861)  loss: 0.7846 (0.7934)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [99]  [ 600/6250]  eta: 0:25:44  lr: 0.000000  grad: 0.2259 (0.3849)  loss: 0.7796 (0.7923)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [99]  [ 700/6250]  eta: 0:25:12  lr: 0.000000  grad: 0.2380 (0.3690)  loss: 0.7848 (0.7913)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [99]  [ 800/6250]  eta: 0:24:41  lr: 0.000000  grad: 0.2552 (0.3783)  loss: 0.7880 (0.7903)  time: 0.2678  data: 0.0001  max mem: 26157
Train: [99]  [ 900/6250]  eta: 0:24:11  lr: 0.000000  grad: 0.2596 (0.3910)  loss: 0.7814 (0.7897)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [99]  [1000/6250]  eta: 0:23:42  lr: 0.000000  grad: 0.2411 (0.3930)  loss: 0.7871 (0.7895)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [99]  [1100/6250]  eta: 0:23:13  lr: 0.000000  grad: 0.2246 (0.3905)  loss: 0.7849 (0.7893)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [1200/6250]  eta: 0:22:44  lr: 0.000000  grad: 0.2519 (0.3828)  loss: 0.7780 (0.7889)  time: 0.2675  data: 0.0001  max mem: 26157
Train: [99]  [1300/6250]  eta: 0:22:16  lr: 0.000000  grad: 0.2502 (0.3828)  loss: 0.7736 (0.7887)  time: 0.2681  data: 0.0002  max mem: 26157
Train: [99]  [1400/6250]  eta: 0:21:48  lr: 0.000000  grad: 0.2556 (0.3848)  loss: 0.7877 (0.7887)  time: 0.2663  data: 0.0001  max mem: 26157
Train: [99]  [1500/6250]  eta: 0:21:20  lr: 0.000000  grad: 0.2308 (0.3801)  loss: 0.7829 (0.7885)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [99]  [1600/6250]  eta: 0:20:52  lr: 0.000000  grad: 0.2236 (0.3728)  loss: 0.7876 (0.7885)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [99]  [1700/6250]  eta: 0:20:25  lr: 0.000000  grad: 0.2251 (0.3672)  loss: 0.7875 (0.7884)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [99]  [1800/6250]  eta: 0:19:57  lr: 0.000000  grad: 0.2383 (0.3754)  loss: 0.7864 (0.7883)  time: 0.2664  data: 0.0002  max mem: 26157
Train: [99]  [1900/6250]  eta: 0:19:30  lr: 0.000000  grad: 0.2181 (0.3740)  loss: 0.7864 (0.7881)  time: 0.2676  data: 0.0002  max mem: 26157
Train: [99]  [2000/6250]  eta: 0:19:02  lr: 0.000000  grad: 0.2120 (0.3740)  loss: 0.7830 (0.7881)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [99]  [2100/6250]  eta: 0:18:35  lr: 0.000000  grad: 0.2318 (0.3685)  loss: 0.7869 (0.7882)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [2200/6250]  eta: 0:18:08  lr: 0.000000  grad: 0.2271 (0.3675)  loss: 0.7791 (0.7881)  time: 0.2677  data: 0.0002  max mem: 26157
Train: [99]  [2300/6250]  eta: 0:17:41  lr: 0.000000  grad: 0.2255 (0.3663)  loss: 0.7826 (0.7881)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [99]  [2400/6250]  eta: 0:17:14  lr: 0.000000  grad: 0.2248 (0.3622)  loss: 0.7849 (0.7880)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [99]  [2500/6250]  eta: 0:16:47  lr: 0.000000  grad: 0.2210 (0.3595)  loss: 0.7916 (0.7881)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [99]  [2600/6250]  eta: 0:16:20  lr: 0.000000  grad: 0.2223 (0.3548)  loss: 0.7889 (0.7882)  time: 0.2657  data: 0.0001  max mem: 26157
Train: [99]  [2700/6250]  eta: 0:15:53  lr: 0.000000  grad: 0.2336 (0.3526)  loss: 0.7878 (0.7881)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [99]  [2800/6250]  eta: 0:15:26  lr: 0.000000  grad: 0.2192 (0.3545)  loss: 0.7918 (0.7881)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [99]  [2900/6250]  eta: 0:14:59  lr: 0.000000  grad: 0.2226 (0.3559)  loss: 0.7820 (0.7881)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [3000/6250]  eta: 0:14:32  lr: 0.000000  grad: 0.2288 (0.3534)  loss: 0.7871 (0.7880)  time: 0.2670  data: 0.0001  max mem: 26157
Train: [99]  [3100/6250]  eta: 0:14:05  lr: 0.000000  grad: 0.2135 (0.3571)  loss: 0.7915 (0.7879)  time: 0.2674  data: 0.0001  max mem: 26157
Train: [99]  [3200/6250]  eta: 0:13:38  lr: 0.000000  grad: 0.2282 (0.3559)  loss: 0.7837 (0.7879)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [99]  [3300/6250]  eta: 0:13:11  lr: 0.000000  grad: 0.2153 (0.3541)  loss: 0.7824 (0.7878)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [99]  [3400/6250]  eta: 0:12:44  lr: 0.000000  grad: 0.2237 (0.3552)  loss: 0.7830 (0.7878)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [99]  [3500/6250]  eta: 0:12:17  lr: 0.000000  grad: 0.2271 (0.3534)  loss: 0.7857 (0.7879)  time: 0.2672  data: 0.0002  max mem: 26157
Train: [99]  [3600/6250]  eta: 0:11:50  lr: 0.000000  grad: 0.2302 (0.3512)  loss: 0.7839 (0.7879)  time: 0.2667  data: 0.0001  max mem: 26157
Train: [99]  [3700/6250]  eta: 0:11:23  lr: 0.000000  grad: 0.2294 (0.3543)  loss: 0.7909 (0.7879)  time: 0.2679  data: 0.0002  max mem: 26157
Train: [99]  [3800/6250]  eta: 0:10:56  lr: 0.000000  grad: 0.2414 (0.3541)  loss: 0.7787 (0.7879)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [99]  [3900/6250]  eta: 0:10:29  lr: 0.000000  grad: 0.2263 (0.3520)  loss: 0.7923 (0.7879)  time: 0.2685  data: 0.0002  max mem: 26157
Train: [99]  [4000/6250]  eta: 0:10:02  lr: 0.000000  grad: 0.2147 (0.3527)  loss: 0.7936 (0.7880)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [4100/6250]  eta: 0:09:36  lr: 0.000000  grad: 0.2123 (0.3508)  loss: 0.7863 (0.7880)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [99]  [4200/6250]  eta: 0:09:09  lr: 0.000000  grad: 0.2194 (0.3501)  loss: 0.7964 (0.7881)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [99]  [4300/6250]  eta: 0:08:42  lr: 0.000000  grad: 0.2241 (0.3485)  loss: 0.7964 (0.7882)  time: 0.2673  data: 0.0002  max mem: 26157
Train: [99]  [4400/6250]  eta: 0:08:15  lr: 0.000000  grad: 0.2099 (0.3478)  loss: 0.8008 (0.7883)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [4500/6250]  eta: 0:07:48  lr: 0.000000  grad: 0.2206 (0.3489)  loss: 0.7873 (0.7884)  time: 0.2664  data: 0.0001  max mem: 26157
Train: [99]  [4600/6250]  eta: 0:07:21  lr: 0.000000  grad: 0.2358 (0.3486)  loss: 0.7905 (0.7884)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [99]  [4700/6250]  eta: 0:06:55  lr: 0.000000  grad: 0.2454 (0.3477)  loss: 0.7845 (0.7884)  time: 0.2669  data: 0.0001  max mem: 26157
Train: [99]  [4800/6250]  eta: 0:06:28  lr: 0.000000  grad: 0.2362 (0.3488)  loss: 0.7894 (0.7884)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [99]  [4900/6250]  eta: 0:06:01  lr: 0.000000  grad: 0.2307 (0.3479)  loss: 0.7848 (0.7884)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [99]  [5000/6250]  eta: 0:05:34  lr: 0.000000  grad: 0.2480 (0.3492)  loss: 0.7874 (0.7883)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [99]  [5100/6250]  eta: 0:05:07  lr: 0.000000  grad: 0.2325 (0.3509)  loss: 0.7858 (0.7884)  time: 0.2677  data: 0.0001  max mem: 26157
Train: [99]  [5200/6250]  eta: 0:04:41  lr: 0.000000  grad: 0.2395 (0.3505)  loss: 0.7911 (0.7884)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [5300/6250]  eta: 0:04:14  lr: 0.000000  grad: 0.2259 (0.3525)  loss: 0.7950 (0.7884)  time: 0.2668  data: 0.0001  max mem: 26157
Train: [99]  [5400/6250]  eta: 0:03:47  lr: 0.000000  grad: 0.2224 (0.3542)  loss: 0.7960 (0.7885)  time: 0.2662  data: 0.0001  max mem: 26157
Train: [99]  [5500/6250]  eta: 0:03:20  lr: 0.000000  grad: 0.2247 (0.3552)  loss: 0.7865 (0.7885)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [5600/6250]  eta: 0:02:53  lr: 0.000000  grad: 0.2480 (0.3557)  loss: 0.7850 (0.7885)  time: 0.2696  data: 0.0002  max mem: 26157
Train: [99]  [5700/6250]  eta: 0:02:27  lr: 0.000000  grad: 0.2208 (0.3551)  loss: 0.7837 (0.7884)  time: 0.2682  data: 0.0001  max mem: 26157
Train: [99]  [5800/6250]  eta: 0:02:00  lr: 0.000000  grad: 0.2342 (0.3559)  loss: 0.7825 (0.7884)  time: 0.2665  data: 0.0001  max mem: 26157
Train: [99]  [5900/6250]  eta: 0:01:33  lr: 0.000000  grad: 0.2352 (0.3561)  loss: 0.7859 (0.7884)  time: 0.2672  data: 0.0001  max mem: 26157
Train: [99]  [6000/6250]  eta: 0:01:06  lr: 0.000000  grad: 0.2309 (0.3556)  loss: 0.7849 (0.7884)  time: 0.2686  data: 0.0001  max mem: 26157
Train: [99]  [6100/6250]  eta: 0:00:40  lr: 0.000000  grad: 0.2386 (0.3552)  loss: 0.7824 (0.7884)  time: 0.2670  data: 0.0002  max mem: 26157
Train: [99]  [6200/6250]  eta: 0:00:13  lr: 0.000000  grad: 0.2191 (0.3540)  loss: 0.7964 (0.7884)  time: 0.2666  data: 0.0001  max mem: 26157
Train: [99]  [6249/6250]  eta: 0:00:00  lr: 0.000000  grad: 0.2338 (0.3564)  loss: 0.7902 (0.7884)  time: 0.2671  data: 0.0001  max mem: 26157
Train: [99] Total time: 0:27:55 (0.2681 s / it)
Averaged stats: lr: 0.000000  grad: 0.2338 (0.3564)  loss: 0.7902 (0.7884)
Eval (hcp-train-subset): [99]  [ 0/62]  eta: 0:03:27  loss: 0.7831 (0.7831)  time: 3.3519  data: 3.2569  max mem: 26157
Eval (hcp-train-subset): [99]  [61/62]  eta: 0:00:00  loss: 0.7774 (0.7837)  time: 0.0926  data: 0.0103  max mem: 26157
Eval (hcp-train-subset): [99] Total time: 0:00:10 (0.1647 s / it)
Averaged stats (hcp-train-subset): loss: 0.7774 (0.7837)
Making plots (hcp-train-subset): example=39
Eval (hcp-val): [99]  [ 0/62]  eta: 0:03:55  loss: 0.8165 (0.8165)  time: 3.8029  data: 3.7199  max mem: 26157
Eval (hcp-val): [99]  [61/62]  eta: 0:00:00  loss: 0.8218 (0.8219)  time: 0.0872  data: 0.0050  max mem: 26157
Eval (hcp-val): [99] Total time: 0:00:10 (0.1661 s / it)
Averaged stats (hcp-val): loss: 0.8218 (0.8219)
Making plots (hcp-val): example=39
saving checkpoint experiments/patch_size/output/patch_size/patch8/pretrain/checkpoint-last.pth
done! training time: 2 days, 0:34:40