Upload 3 files

The Res-VMamba weight in paper https://arxiv.org/abs/2402.15761 , which was trained on CNFOOD-241-Chen.

ckpt_epoch_166.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:457b615d41c79698d8f2eafbe51959d8c1b5d53187605765d5f79558639c1ac3
+size 711402283

config.json

@@ -0,0 +1,99 @@
+AMP_ENABLE: true
+AMP_OPT_LEVEL: ''
+AUG:
+  AUTO_AUGMENT: rand-m9-mstd0.5-inc1
+  COLOR_JITTER: 0.4
+  CUTMIX: 1.0
+  CUTMIX_MINMAX: null
+  MIXUP: 0.8
+  MIXUP_MODE: batch
+  MIXUP_PROB: 1.0
+  MIXUP_SWITCH_PROB: 0.5
+  RECOUNT: 1
+  REMODE: pixel
+  REPROB: 0.25
+BASE:
+- ''
+DATA:
+  BATCH_SIZE: 128
+  CACHE_MODE: part
+  DATASET: imagenet
+  DATA_PATH: /home/public_3T/food_data/CNFOOD-241
+  IMG_SIZE: 224
+  INTERPOLATION: bicubic
+  MASK_PATCH_SIZE: 32
+  MASK_RATIO: 0.6
+  NUM_WORKERS: 8
+  PIN_MEMORY: true
+  ZIP_MODE: false
+ENABLE_AMP: false
+EVAL_MODE: false
+FUSED_LAYERNORM: false
+FUSED_WINDOW_PROCESS: false
+LOCAL_RANK: 0
+MODEL:
+  DROP_PATH_RATE: 0.3
+  DROP_RATE: 0.0
+  LABEL_SMOOTHING: 0.1
+  MMCKPT: false
+  NAME: vssm_small
+  NUM_CLASSES: 241
+  PRETRAINED: ./res_vmamba_cnf241_result_2/vssm_small/default/ckpt_epoch_12.pth
+  RESUME: ''
+  TYPE: vssm
+  VSSM:
+    DEPTHS:
+    - 2
+    - 2
+    - 27
+    - 2
+    DOWNSAMPLE: v1
+    DT_RANK: auto
+    D_STATE: 16
+    EMBED_DIM: 96
+    IN_CHANS: 3
+    MLP_RATIO: 0.0
+    PATCH_NORM: true
+    PATCH_SIZE: 4
+    SHARED_SSM: false
+    SOFTMAX: false
+    SSM_RATIO: 2.0
+OUTPUT: ./res_vmamba_cnf241_result_best/vssm_small/default
+PRINT_FREQ: 10
+SAVE_FREQ: 1
+SEED: 0
+TAG: default
+TEST:
+  CROP: true
+  SEQUENTIAL: false
+  SHUFFLE: false
+THROUGHPUT_MODE: false
+TRAIN:
+  ACCUMULATION_STEPS: 1
+  AUTO_RESUME: true
+  BASE_LR: 0.000125
+  CLIP_GRAD: 5.0
+  EPOCHS: 300
+  LAYER_DECAY: 1.0
+  LR_SCHEDULER:
+    DECAY_EPOCHS: 30
+    DECAY_RATE: 0.1
+    GAMMA: 0.1
+    MULTISTEPS: []
+    NAME: cosine
+    WARMUP_PREFIX: true
+  MIN_LR: 1.25e-06
+  MOE:
+    SAVE_MASTER: false
+  OPTIMIZER:
+    BETAS:
+    - 0.9
+    - 0.999
+    EPS: 1.0e-08
+    MOMENTUM: 0.9
+    NAME: adamw
+  START_EPOCH: 0
+  USE_CHECKPOINT: false
+  WARMUP_EPOCHS: 20
+  WARMUP_LR: 1.25e-07
+  WEIGHT_DECAY: 0.05

log_rank0.txt

@@ -0,0 +1,1233 @@
+[2024-02-22 17:55:19 vssm_small] (main.py 401): INFO Full config saved to ./res_vmamba_cnf241_result_best/vssm_small/default/config.json
+[2024-02-22 17:55:19 vssm_small] (main.py 404): INFO AMP_ENABLE: true
+AMP_OPT_LEVEL: ''
+AUG:
+  AUTO_AUGMENT: rand-m9-mstd0.5-inc1
+  COLOR_JITTER: 0.4
+  CUTMIX: 1.0
+  CUTMIX_MINMAX: null
+  MIXUP: 0.8
+  MIXUP_MODE: batch
+  MIXUP_PROB: 1.0
+  MIXUP_SWITCH_PROB: 0.5
+  RECOUNT: 1
+  REMODE: pixel
+  REPROB: 0.25
+BASE:
+- ''
+DATA:
+  BATCH_SIZE: 128
+  CACHE_MODE: part
+  DATASET: imagenet
+  DATA_PATH: /home/public_3T/food_data/CNFOOD-241
+  IMG_SIZE: 224
+  INTERPOLATION: bicubic
+  MASK_PATCH_SIZE: 32
+  MASK_RATIO: 0.6
+  NUM_WORKERS: 8
+  PIN_MEMORY: true
+  ZIP_MODE: false
+ENABLE_AMP: false
+EVAL_MODE: false
+FUSED_LAYERNORM: false
+FUSED_WINDOW_PROCESS: false
+LOCAL_RANK: 0
+MODEL:
+  DROP_PATH_RATE: 0.3
+  DROP_RATE: 0.0
+  LABEL_SMOOTHING: 0.1
+  MMCKPT: false
+  NAME: vssm_small
+  NUM_CLASSES: 241
+  PRETRAINED: ./res_vmamba_cnf241_result_2/vssm_small/default/ckpt_epoch_12.pth
+  RESUME: ''
+  TYPE: vssm
+  VSSM:
+    DEPTHS:
+    - 2
+    - 2
+    - 27
+    - 2
+    DOWNSAMPLE: v1
+    DT_RANK: auto
+    D_STATE: 16
+    EMBED_DIM: 96
+    IN_CHANS: 3
+    MLP_RATIO: 0.0
+    PATCH_NORM: true
+    PATCH_SIZE: 4
+    SHARED_SSM: false
+    SOFTMAX: false
+    SSM_RATIO: 2.0
+OUTPUT: ./res_vmamba_cnf241_result_best/vssm_small/default
+PRINT_FREQ: 10
+SAVE_FREQ: 1
+SEED: 0
+TAG: default
+TEST:
+  CROP: true
+  SEQUENTIAL: false
+  SHUFFLE: false
+THROUGHPUT_MODE: false
+TRAIN:
+  ACCUMULATION_STEPS: 1
+  AUTO_RESUME: true
+  BASE_LR: 0.000125
+  CLIP_GRAD: 5.0
+  EPOCHS: 300
+  LAYER_DECAY: 1.0
+  LR_SCHEDULER:
+    DECAY_EPOCHS: 30
+    DECAY_RATE: 0.1
+    GAMMA: 0.1
+    MULTISTEPS: []
+    NAME: cosine
+    WARMUP_PREFIX: true
+  MIN_LR: 1.25e-06
+  MOE:
+    SAVE_MASTER: false
+  OPTIMIZER:
+    BETAS:
+    - 0.9
+    - 0.999
+    EPS: 1.0e-08
+    MOMENTUM: 0.9
+    NAME: adamw
+  START_EPOCH: 0
+  USE_CHECKPOINT: false
+  WARMUP_EPOCHS: 20
+  WARMUP_LR: 1.25e-07
+  WEIGHT_DECAY: 0.05
+
+[2024-02-22 17:55:19 vssm_small] (main.py 405): INFO {"cfg": "configs/vssm/vssm_small_224.yaml", "opts": null, "batch_size": 128, "data_path": "/home/public_3T/food_data/CNFOOD-241", "zip": false, "cache_mode": "part", "pretrained": "./res_vmamba_cnf241_result_2/vssm_small/default/ckpt_epoch_12.pth", "resume": null, "accumulation_steps": null, "use_checkpoint": false, "disable_amp": false, "amp_opt_level": null, "output": "./res_vmamba_cnf241_result_best", "tag": null, "eval": false, "throughput": false, "local_rank": 0, "fused_layernorm": false, "optim": null, "model_ema": true, "model_ema_decay": 0.9999, "model_ema_force_cpu": false}
+[2024-02-22 17:55:20 vssm_small] (main.py 112): INFO Creating model:vssm/vssm_small
+[2024-02-22 17:55:20 vssm_small] (main.py 118): INFO VSSM(
+  (patch_embed): Sequential(
+    (0): Conv2d(3, 96, kernel_size=(4, 4), stride=(4, 4))
+    (1): Permute()
+    (2): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
+  )
+  (layers): ModuleList(
+    (0): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=96, out_features=384, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(192, 192, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=192)
+            (out_proj): Linear(in_features=192, out_features=96, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.0)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=96, out_features=384, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(192, 192, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=192)
+            (out_proj): Linear(in_features=192, out_features=96, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.00937500037252903)
+        )
+      )
+      (downsample): PatchMerging2D(
+        (reduction): Linear(in_features=384, out_features=192, bias=False)
+        (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+      )
+    )
+    (1): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=192, out_features=768, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384)
+            (out_proj): Linear(in_features=384, out_features=192, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.01875000074505806)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=192, out_features=768, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384)
+            (out_proj): Linear(in_features=384, out_features=192, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.02812500111758709)
+        )
+      )
+      (downsample): PatchMerging2D(
+        (reduction): Linear(in_features=768, out_features=384, bias=False)
+        (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+      )
+    )
+    (2): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.03750000149011612)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.046875)
+        )
+        (2): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.05625000223517418)
+        )
+        (3): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.06562500447034836)
+        )
+        (4): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.07500000298023224)
+        )
+        (5): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.08437500149011612)
+        )
+        (6): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.09375)
+        )
+        (7): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.10312500596046448)
+        )
+        (8): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.11250000447034836)
+        )
+        (9): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.12187500298023224)
+        )
+        (10): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.13125000894069672)
+        )
+        (11): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.140625)
+        )
+        (12): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.15000000596046448)
+        )
+        (13): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.15937501192092896)
+        )
+        (14): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.16875000298023224)
+        )
+        (15): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.17812500894069672)
+        )
+        (16): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.1875)
+        )
+        (17): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.19687500596046448)
+        )
+        (18): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.20625001192092896)
+        )
+        (19): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.21562501788139343)
+        )
+        (20): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.22500000894069672)
+        )
+        (21): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.2343750149011612)
+        )
+        (22): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.24375000596046448)
+        )
+        (23): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.25312501192092896)
+        )
+        (24): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.26250001788139343)
+        )
+        (25): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.2718750238418579)
+        )
+        (26): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.28125)
+        )
+      )
+      (downsample): PatchMerging2D(
+        (reduction): Linear(in_features=1536, out_features=768, bias=False)
+        (norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
+      )
+    )
+    (3): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=768, out_features=3072, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(1536, 1536, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1536)
+            (out_proj): Linear(in_features=1536, out_features=768, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.2906250059604645)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=768, out_features=3072, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(1536, 1536, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1536)
+            (out_proj): Linear(in_features=1536, out_features=768, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.30000001192092896)
+        )
+      )
+      (downsample): Identity()
+    )
+  )
+  (classifier): Sequential(
+    (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+    (permute): Permute()
+    (avgpool): AdaptiveAvgPool2d(output_size=1)
+    (flatten): Flatten(start_dim=1, end_dim=-1)
+    (head): Linear(in_features=768, out_features=1000, bias=True)
+  )
+)
+[2024-02-22 17:55:20 vssm_small] (main.py 120): INFO number of params: 44417416
+[2024-02-22 17:55:22 vssm_small] (main.py 123): INFO number of GFLOPs: 11.231522784
+[2024-02-22 17:55:22 vssm_small] (main.py 167): INFO auto resuming from ./res_vmamba_cnf241_result_best/vssm_small/default/ckpt_epoch_166.pth
+[2024-02-22 17:55:22 vssm_small] (utils.py 18): INFO ==============> Resuming form ./res_vmamba_cnf241_result_best/vssm_small/default/ckpt_epoch_166.pth....................
+[2024-02-22 17:55:23 vssm_small] (utils.py 27): INFO resuming model: <All keys matched successfully>
+[2024-02-22 17:55:23 vssm_small] (utils.py 34): INFO resuming model_ema: <All keys matched successfully>
+[2024-02-22 17:55:24 vssm_small] (utils.py 48): INFO => loaded successfully './res_vmamba_cnf241_result_best/vssm_small/default/ckpt_epoch_166.pth' (epoch 166)
+[2024-02-22 17:55:34 vssm_small] (main.py 324): INFO Test: [0/402]	Time 10.582 (10.582)	Loss 0.4614 (0.4614)	Acc@1 89.844 (89.844)	Acc@5 97.656 (97.656)	Mem 7155MB
+[2024-02-22 17:55:39 vssm_small] (main.py 324): INFO Test: [10/402]	Time 0.483 (1.401)	Loss 1.1680 (0.9504)	Acc@1 76.562 (80.114)	Acc@5 92.188 (94.105)	Mem 7155MB
+[2024-02-22 17:55:44 vssm_small] (main.py 324): INFO Test: [20/402]	Time 0.482 (0.964)	Loss 1.6357 (0.9252)	Acc@1 52.344 (78.720)	Acc@5 92.188 (95.164)	Mem 7155MB
+[2024-02-22 17:55:49 vssm_small] (main.py 324): INFO Test: [30/402]	Time 0.483 (0.809)	Loss 0.9429 (0.9577)	Acc@1 76.562 (77.848)	Acc@5 98.438 (95.237)	Mem 7155MB
+[2024-02-22 17:55:53 vssm_small] (main.py 324): INFO Test: [40/402]	Time 0.483 (0.729)	Loss 1.0166 (0.9836)	Acc@1 72.656 (76.791)	Acc@5 96.875 (95.560)	Mem 7155MB
+[2024-02-22 17:55:58 vssm_small] (main.py 324): INFO Test: [50/402]	Time 0.483 (0.681)	Loss 0.6353 (0.9501)	Acc@1 83.594 (77.788)	Acc@5 97.656 (95.527)	Mem 7155MB
+[2024-02-22 17:56:03 vssm_small] (main.py 324): INFO Test: [60/402]	Time 0.482 (0.648)	Loss 0.7671 (0.9817)	Acc@1 80.469 (77.011)	Acc@5 96.875 (95.197)	Mem 7155MB
+[2024-02-22 17:56:08 vssm_small] (main.py 324): INFO Test: [70/402]	Time 0.482 (0.625)	Loss 0.8667 (0.9259)	Acc@1 77.344 (78.444)	Acc@5 96.094 (95.478)	Mem 7155MB
+[2024-02-22 17:56:13 vssm_small] (main.py 324): INFO Test: [80/402]	Time 0.483 (0.607)	Loss 0.4990 (0.9246)	Acc@1 88.281 (78.511)	Acc@5 99.219 (95.515)	Mem 7155MB
+[2024-02-22 17:56:18 vssm_small] (main.py 324): INFO Test: [90/402]	Time 0.483 (0.594)	Loss 0.3621 (0.8928)	Acc@1 92.188 (79.327)	Acc@5 100.000 (95.639)	Mem 7155MB
+[2024-02-22 17:56:22 vssm_small] (main.py 324): INFO Test: [100/402]	Time 0.482 (0.583)	Loss 1.1924 (0.9116)	Acc@1 78.125 (79.038)	Acc@5 90.625 (95.359)	Mem 7155MB
+[2024-02-22 17:56:27 vssm_small] (main.py 324): INFO Test: [110/402]	Time 0.483 (0.574)	Loss 0.4204 (0.9173)	Acc@1 85.938 (78.899)	Acc@5 99.219 (95.341)	Mem 7155MB
+[2024-02-22 17:56:32 vssm_small] (main.py 324): INFO Test: [120/402]	Time 0.483 (0.566)	Loss 2.3027 (0.9299)	Acc@1 46.094 (78.622)	Acc@5 83.594 (95.274)	Mem 7155MB
+[2024-02-22 17:56:37 vssm_small] (main.py 324): INFO Test: [130/402]	Time 0.482 (0.560)	Loss 0.4097 (0.9233)	Acc@1 91.406 (78.715)	Acc@5 99.219 (95.366)	Mem 7155MB
+[2024-02-22 17:56:42 vssm_small] (main.py 324): INFO Test: [140/402]	Time 0.483 (0.554)	Loss 2.3652 (0.9762)	Acc@1 47.656 (77.543)	Acc@5 75.781 (94.891)	Mem 7155MB
+[2024-02-22 17:56:47 vssm_small] (main.py 324): INFO Test: [150/402]	Time 0.482 (0.549)	Loss 0.9092 (0.9862)	Acc@1 78.906 (77.266)	Acc@5 93.750 (94.868)	Mem 7155MB
+[2024-02-22 17:56:51 vssm_small] (main.py 324): INFO Test: [160/402]	Time 0.482 (0.545)	Loss 1.4434 (0.9815)	Acc@1 61.719 (77.300)	Acc@5 93.750 (94.978)	Mem 7155MB
+[2024-02-22 17:56:56 vssm_small] (main.py 324): INFO Test: [170/402]	Time 0.483 (0.542)	Loss 0.6587 (0.9758)	Acc@1 89.062 (77.421)	Acc@5 94.531 (95.006)	Mem 7155MB
+[2024-02-22 17:57:01 vssm_small] (main.py 324): INFO Test: [180/402]	Time 0.482 (0.538)	Loss 1.8477 (0.9902)	Acc@1 53.125 (77.175)	Acc@5 93.750 (94.864)	Mem 7155MB
+[2024-02-22 17:57:06 vssm_small] (main.py 324): INFO Test: [190/402]	Time 0.483 (0.535)	Loss 0.4958 (0.9845)	Acc@1 89.844 (77.356)	Acc@5 96.875 (94.818)	Mem 7155MB
+[2024-02-22 17:57:11 vssm_small] (main.py 324): INFO Test: [200/402]	Time 0.482 (0.533)	Loss 0.3074 (0.9707)	Acc@1 95.312 (77.697)	Acc@5 97.656 (94.916)	Mem 7155MB
+[2024-02-22 17:57:16 vssm_small] (main.py 324): INFO Test: [210/402]	Time 0.483 (0.530)	Loss 0.5928 (0.9563)	Acc@1 83.594 (78.058)	Acc@5 99.219 (95.016)	Mem 7155MB
+[2024-02-22 17:57:20 vssm_small] (main.py 324): INFO Test: [220/402]	Time 0.482 (0.528)	Loss 1.1055 (0.9381)	Acc@1 76.562 (78.450)	Acc@5 92.969 (95.178)	Mem 7155MB
+[2024-02-22 17:57:25 vssm_small] (main.py 324): INFO Test: [230/402]	Time 0.482 (0.526)	Loss 2.1230 (0.9502)	Acc@1 60.156 (78.436)	Acc@5 78.906 (94.913)	Mem 7155MB
+[2024-02-22 17:57:30 vssm_small] (main.py 324): INFO Test: [240/402]	Time 0.483 (0.524)	Loss 1.1201 (0.9431)	Acc@1 67.188 (78.618)	Acc@5 98.438 (94.972)	Mem 7155MB
+[2024-02-22 17:57:35 vssm_small] (main.py 324): INFO Test: [250/402]	Time 0.483 (0.523)	Loss 1.8711 (0.9650)	Acc@1 53.906 (78.019)	Acc@5 95.312 (94.933)	Mem 7155MB
+[2024-02-22 17:57:40 vssm_small] (main.py 324): INFO Test: [260/402]	Time 0.482 (0.521)	Loss 0.9282 (0.9637)	Acc@1 76.562 (78.023)	Acc@5 99.219 (94.983)	Mem 7155MB
+[2024-02-22 17:57:44 vssm_small] (main.py 324): INFO Test: [270/402]	Time 0.483 (0.520)	Loss 1.1191 (0.9527)	Acc@1 68.750 (78.269)	Acc@5 94.531 (95.056)	Mem 7155MB
+[2024-02-22 17:57:49 vssm_small] (main.py 324): INFO Test: [280/402]	Time 0.483 (0.519)	Loss 2.3047 (0.9652)	Acc@1 19.531 (77.755)	Acc@5 92.188 (95.032)	Mem 7155MB
+[2024-02-22 17:57:54 vssm_small] (main.py 324): INFO Test: [290/402]	Time 0.482 (0.517)	Loss 0.8774 (0.9767)	Acc@1 80.469 (77.489)	Acc@5 93.750 (94.912)	Mem 7155MB
+[2024-02-22 17:57:59 vssm_small] (main.py 324): INFO Test: [300/402]	Time 0.483 (0.516)	Loss 1.0645 (0.9802)	Acc@1 80.469 (77.512)	Acc@5 92.188 (94.817)	Mem 7155MB
+[2024-02-22 17:58:04 vssm_small] (main.py 324): INFO Test: [310/402]	Time 0.483 (0.515)	Loss 1.0410 (0.9764)	Acc@1 79.688 (77.462)	Acc@5 96.094 (94.903)	Mem 7155MB
+[2024-02-22 17:58:09 vssm_small] (main.py 324): INFO Test: [320/402]	Time 0.483 (0.514)	Loss 1.0859 (0.9653)	Acc@1 76.562 (77.743)	Acc@5 89.844 (94.960)	Mem 7155MB
+[2024-02-22 17:58:13 vssm_small] (main.py 324): INFO Test: [330/402]	Time 0.483 (0.513)	Loss 1.0596 (0.9657)	Acc@1 73.438 (77.714)	Acc@5 95.312 (95.001)	Mem 7155MB
+[2024-02-22 17:58:18 vssm_small] (main.py 324): INFO Test: [340/402]	Time 0.482 (0.512)	Loss 0.3967 (0.9663)	Acc@1 90.625 (77.699)	Acc@5 100.000 (95.028)	Mem 7155MB
+[2024-02-22 17:58:23 vssm_small] (main.py 324): INFO Test: [350/402]	Time 0.483 (0.511)	Loss 1.2148 (0.9637)	Acc@1 68.750 (77.773)	Acc@5 96.875 (95.050)	Mem 7155MB
+[2024-02-22 17:58:28 vssm_small] (main.py 324): INFO Test: [360/402]	Time 0.483 (0.511)	Loss 0.9941 (0.9685)	Acc@1 79.688 (77.571)	Acc@5 95.312 (95.074)	Mem 7155MB
+[2024-02-22 17:58:33 vssm_small] (main.py 324): INFO Test: [370/402]	Time 0.482 (0.510)	Loss 0.9004 (0.9689)	Acc@1 83.594 (77.552)	Acc@5 94.531 (95.081)	Mem 7155MB
+[2024-02-22 17:58:38 vssm_small] (main.py 324): INFO Test: [380/402]	Time 0.482 (0.509)	Loss 0.7358 (0.9634)	Acc@1 82.812 (77.690)	Acc@5 97.656 (95.114)	Mem 7155MB
+[2024-02-22 17:58:42 vssm_small] (main.py 324): INFO Test: [390/402]	Time 0.482 (0.508)	Loss 1.0068 (0.9605)	Acc@1 77.344 (77.807)	Acc@5 94.531 (95.113)	Mem 7155MB
+[2024-02-22 17:58:47 vssm_small] (main.py 324): INFO Test: [400/402]	Time 0.482 (0.508)	Loss 0.2834 (0.9484)	Acc@1 95.312 (78.141)	Acc@5 99.219 (95.184)	Mem 7155MB
+[2024-02-22 17:58:48 vssm_small] (main.py 331): INFO  * Acc@1 78.150 Acc@5 95.186
+[2024-02-22 17:58:48 vssm_small] (main.py 174): INFO Accuracy of the network on the 51354 test images: 78.1%
+[2024-02-22 17:58:57 vssm_small] (main.py 324): INFO Test: [0/402]	Time 8.919 (8.919)	Loss 0.4504 (0.4504)	Acc@1 91.406 (91.406)	Acc@5 98.438 (98.438)	Mem 7155MB
+[2024-02-22 17:59:02 vssm_small] (main.py 324): INFO Test: [10/402]	Time 0.483 (1.250)	Loss 0.9731 (0.8429)	Acc@1 83.594 (82.599)	Acc@5 91.406 (94.957)	Mem 7155MB
+[2024-02-22 17:59:07 vssm_small] (main.py 324): INFO Test: [20/402]	Time 0.483 (0.884)	Loss 1.3154 (0.8200)	Acc@1 60.938 (81.510)	Acc@5 94.531 (95.573)	Mem 7155MB
+[2024-02-22 17:59:12 vssm_small] (main.py 324): INFO Test: [30/402]	Time 0.483 (0.755)	Loss 0.8833 (0.8867)	Acc@1 76.562 (79.410)	Acc@5 96.875 (95.514)	Mem 7155MB
+[2024-02-22 17:59:16 vssm_small] (main.py 324): INFO Test: [40/402]	Time 0.482 (0.688)	Loss 0.8809 (0.8790)	Acc@1 74.219 (79.002)	Acc@5 98.438 (95.922)	Mem 7155MB
+[2024-02-22 17:59:21 vssm_small] (main.py 324): INFO Test: [50/402]	Time 0.483 (0.648)	Loss 0.5254 (0.8631)	Acc@1 89.062 (79.611)	Acc@5 97.656 (95.956)	Mem 7155MB
+[2024-02-22 17:59:26 vssm_small] (main.py 324): INFO Test: [60/402]	Time 0.483 (0.621)	Loss 0.6147 (0.8904)	Acc@1 85.156 (78.855)	Acc@5 97.656 (95.671)	Mem 7155MB
+[2024-02-22 17:59:31 vssm_small] (main.py 324): INFO Test: [70/402]	Time 0.483 (0.601)	Loss 1.0029 (0.8448)	Acc@1 77.344 (80.095)	Acc@5 96.875 (96.006)	Mem 7155MB
+[2024-02-22 17:59:36 vssm_small] (main.py 324): INFO Test: [80/402]	Time 0.483 (0.587)	Loss 0.5259 (0.8449)	Acc@1 85.156 (80.102)	Acc@5 98.438 (96.007)	Mem 7155MB
+[2024-02-22 17:59:41 vssm_small] (main.py 324): INFO Test: [90/402]	Time 0.483 (0.575)	Loss 0.2947 (0.8155)	Acc@1 94.531 (80.872)	Acc@5 100.000 (96.162)	Mem 7155MB
+[2024-02-22 17:59:45 vssm_small] (main.py 324): INFO Test: [100/402]	Time 0.483 (0.566)	Loss 1.2002 (0.8335)	Acc@1 76.562 (80.554)	Acc@5 92.188 (95.978)	Mem 7155MB
+[2024-02-22 17:59:50 vssm_small] (main.py 324): INFO Test: [110/402]	Time 0.483 (0.559)	Loss 0.4329 (0.8417)	Acc@1 86.719 (80.342)	Acc@5 100.000 (95.967)	Mem 7155MB
+[2024-02-22 17:59:55 vssm_small] (main.py 324): INFO Test: [120/402]	Time 0.483 (0.552)	Loss 2.2422 (0.8554)	Acc@1 47.656 (80.139)	Acc@5 84.375 (95.874)	Mem 7155MB
+[2024-02-22 18:00:00 vssm_small] (main.py 324): INFO Test: [130/402]	Time 0.483 (0.547)	Loss 0.4048 (0.8500)	Acc@1 90.625 (80.200)	Acc@5 99.219 (95.974)	Mem 7155MB
+[2024-02-22 18:00:05 vssm_small] (main.py 324): INFO Test: [140/402]	Time 0.483 (0.543)	Loss 2.1191 (0.9016)	Acc@1 52.344 (79.039)	Acc@5 83.594 (95.495)	Mem 7155MB
+[2024-02-22 18:00:09 vssm_small] (main.py 324): INFO Test: [150/402]	Time 0.483 (0.539)	Loss 0.8765 (0.9130)	Acc@1 78.906 (78.715)	Acc@5 94.531 (95.442)	Mem 7155MB
+[2024-02-22 18:00:14 vssm_small] (main.py 324): INFO Test: [160/402]	Time 0.483 (0.535)	Loss 1.3135 (0.9056)	Acc@1 67.969 (78.872)	Acc@5 95.312 (95.541)	Mem 7155MB
+[2024-02-22 18:00:19 vssm_small] (main.py 324): INFO Test: [170/402]	Time 0.483 (0.532)	Loss 0.5923 (0.8953)	Acc@1 90.625 (79.094)	Acc@5 95.312 (95.591)	Mem 7155MB
+[2024-02-22 18:00:24 vssm_small] (main.py 324): INFO Test: [180/402]	Time 0.483 (0.529)	Loss 1.8027 (0.9146)	Acc@1 53.125 (78.699)	Acc@5 94.531 (95.395)	Mem 7155MB
+[2024-02-22 18:00:29 vssm_small] (main.py 324): INFO Test: [190/402]	Time 0.483 (0.527)	Loss 0.4436 (0.9099)	Acc@1 91.406 (78.865)	Acc@5 97.656 (95.357)	Mem 7155MB
+[2024-02-22 18:00:34 vssm_small] (main.py 324): INFO Test: [200/402]	Time 0.483 (0.525)	Loss 0.2937 (0.8963)	Acc@1 96.875 (79.190)	Acc@5 98.438 (95.464)	Mem 7155MB
+[2024-02-22 18:00:38 vssm_small] (main.py 324): INFO Test: [210/402]	Time 0.483 (0.523)	Loss 0.5981 (0.8853)	Acc@1 84.375 (79.465)	Acc@5 98.438 (95.542)	Mem 7155MB
+[2024-02-22 18:00:43 vssm_small] (main.py 324): INFO Test: [220/402]	Time 0.483 (0.521)	Loss 1.0889 (0.8694)	Acc@1 77.344 (79.811)	Acc@5 92.969 (95.680)	Mem 7155MB
+[2024-02-22 18:00:48 vssm_small] (main.py 324): INFO Test: [230/402]	Time 0.483 (0.519)	Loss 1.9727 (0.8842)	Acc@1 60.156 (79.708)	Acc@5 78.906 (95.394)	Mem 7155MB
+[2024-02-22 18:00:53 vssm_small] (main.py 324): INFO Test: [240/402]	Time 0.482 (0.518)	Loss 1.2422 (0.8778)	Acc@1 60.156 (79.853)	Acc@5 98.438 (95.445)	Mem 7155MB
+[2024-02-22 18:00:58 vssm_small] (main.py 324): INFO Test: [250/402]	Time 0.483 (0.516)	Loss 1.4551 (0.8951)	Acc@1 60.938 (79.358)	Acc@5 95.312 (95.412)	Mem 7155MB
+[2024-02-22 18:01:03 vssm_small] (main.py 324): INFO Test: [260/402]	Time 0.483 (0.515)	Loss 0.8667 (0.8933)	Acc@1 78.906 (79.331)	Acc@5 98.438 (95.489)	Mem 7155MB
+[2024-02-22 18:01:07 vssm_small] (main.py 324): INFO Test: [270/402]	Time 0.482 (0.514)	Loss 0.9072 (0.8828)	Acc@1 77.344 (79.578)	Acc@5 96.094 (95.543)	Mem 7155MB
+[2024-02-22 18:01:12 vssm_small] (main.py 324): INFO Test: [280/402]	Time 0.483 (0.513)	Loss 2.3594 (0.8950)	Acc@1 19.531 (79.101)	Acc@5 92.188 (95.527)	Mem 7155MB
+[2024-02-22 18:01:17 vssm_small] (main.py 324): INFO Test: [290/402]	Time 0.483 (0.512)	Loss 0.8384 (0.9058)	Acc@1 82.031 (78.845)	Acc@5 95.312 (95.455)	Mem 7155MB
+[2024-02-22 18:01:22 vssm_small] (main.py 324): INFO Test: [300/402]	Time 0.482 (0.511)	Loss 0.9658 (0.9067)	Acc@1 81.250 (78.914)	Acc@5 93.750 (95.396)	Mem 7155MB
+[2024-02-22 18:01:27 vssm_small] (main.py 324): INFO Test: [310/402]	Time 0.483 (0.510)	Loss 1.0488 (0.9032)	Acc@1 81.250 (78.861)	Acc@5 96.094 (95.481)	Mem 7155MB
+[2024-02-22 18:01:32 vssm_small] (main.py 324): INFO Test: [320/402]	Time 0.483 (0.509)	Loss 0.8892 (0.8902)	Acc@1 82.031 (79.193)	Acc@5 92.188 (95.536)	Mem 7155MB
+[2024-02-22 18:01:36 vssm_small] (main.py 324): INFO Test: [330/402]	Time 0.483 (0.508)	Loss 0.8677 (0.8919)	Acc@1 79.688 (79.145)	Acc@5 97.656 (95.567)	Mem 7155MB
+[2024-02-22 18:01:41 vssm_small] (main.py 324): INFO Test: [340/402]	Time 0.483 (0.507)	Loss 0.3433 (0.8911)	Acc@1 89.844 (79.138)	Acc@5 100.000 (95.597)	Mem 7155MB
+[2024-02-22 18:01:46 vssm_small] (main.py 324): INFO Test: [350/402]	Time 0.483 (0.507)	Loss 0.8315 (0.8892)	Acc@1 78.906 (79.200)	Acc@5 98.438 (95.620)	Mem 7155MB
+[2024-02-22 18:01:51 vssm_small] (main.py 324): INFO Test: [360/402]	Time 0.483 (0.506)	Loss 0.9419 (0.8932)	Acc@1 78.125 (79.006)	Acc@5 96.094 (95.654)	Mem 7155MB
+[2024-02-22 18:01:56 vssm_small] (main.py 324): INFO Test: [370/402]	Time 0.483 (0.505)	Loss 0.8735 (0.8931)	Acc@1 82.031 (78.997)	Acc@5 94.531 (95.652)	Mem 7155MB
+[2024-02-22 18:02:01 vssm_small] (main.py 324): INFO Test: [380/402]	Time 0.483 (0.505)	Loss 0.7627 (0.8883)	Acc@1 82.031 (79.138)	Acc@5 97.656 (95.682)	Mem 7155MB
+[2024-02-22 18:02:05 vssm_small] (main.py 324): INFO Test: [390/402]	Time 0.483 (0.504)	Loss 0.9995 (0.8870)	Acc@1 78.125 (79.218)	Acc@5 93.750 (95.666)	Mem 7155MB
+[2024-02-22 18:02:10 vssm_small] (main.py 324): INFO Test: [400/402]	Time 0.482 (0.504)	Loss 0.2445 (0.8753)	Acc@1 96.094 (79.534)	Acc@5 99.219 (95.722)	Mem 7155MB
+[2024-02-22 18:02:11 vssm_small] (main.py 331): INFO  * Acc@1 79.544 Acc@5 95.724
+[2024-02-22 18:02:11 vssm_small] (main.py 177): INFO Accuracy of the network ema on the 51354 test images: 79.5%
+[2024-02-22 18:02:11 vssm_small] (main.py 196): INFO Start training
+[2024-02-22 18:02:22 vssm_small] (main.py 274): INFO Train: [167/300][0/933]	eta 2:59:28 lr 0.000058	 wd 0.0500	time 11.5413 (11.5413)	data time 8.5294 (8.5294)	loss 3.2837 (3.2837)	grad_norm 7.2747 (7.2747)	loss_scale 32768.0000 (32768.0000)	mem 50097MB
+[2024-02-22 18:02:38 vssm_small] (main.py 274): INFO Train: [167/300][10/933]	eta 0:38:04 lr 0.000058	 wd 0.0500	time 1.5679 (2.4753)	data time 0.0005 (0.7759)	loss 2.0028 (3.0696)	grad_norm 5.6496 (6.8319)	loss_scale 32768.0000 (32768.0000)	mem 50285MB
+[2024-02-22 18:03:56 vssm_small] (main.py 401): INFO Full config saved to ./res_vmamba_cnf241_result_best/vssm_small/default/config.json
+[2024-02-22 18:03:56 vssm_small] (main.py 404): INFO AMP_ENABLE: true
+AMP_OPT_LEVEL: ''
+AUG:
+  AUTO_AUGMENT: rand-m9-mstd0.5-inc1
+  COLOR_JITTER: 0.4
+  CUTMIX: 1.0
+  CUTMIX_MINMAX: null
+  MIXUP: 0.8
+  MIXUP_MODE: batch
+  MIXUP_PROB: 1.0
+  MIXUP_SWITCH_PROB: 0.5
+  RECOUNT: 1
+  REMODE: pixel
+  REPROB: 0.25
+BASE:
+- ''
+DATA:
+  BATCH_SIZE: 128
+  CACHE_MODE: part
+  DATASET: imagenet
+  DATA_PATH: /home/public_3T/food_data/CNFOOD-241
+  IMG_SIZE: 224
+  INTERPOLATION: bicubic
+  MASK_PATCH_SIZE: 32
+  MASK_RATIO: 0.6
+  NUM_WORKERS: 8
+  PIN_MEMORY: true
+  ZIP_MODE: false
+ENABLE_AMP: false
+EVAL_MODE: false
+FUSED_LAYERNORM: false
+FUSED_WINDOW_PROCESS: false
+LOCAL_RANK: 0
+MODEL:
+  DROP_PATH_RATE: 0.3
+  DROP_RATE: 0.0
+  LABEL_SMOOTHING: 0.1
+  MMCKPT: false
+  NAME: vssm_small
+  NUM_CLASSES: 241
+  PRETRAINED: ./res_vmamba_cnf241_result_2/vssm_small/default/ckpt_epoch_12.pth
+  RESUME: ''
+  TYPE: vssm
+  VSSM:
+    DEPTHS:
+    - 2
+    - 2
+    - 27
+    - 2
+    DOWNSAMPLE: v1
+    DT_RANK: auto
+    D_STATE: 16
+    EMBED_DIM: 96
+    IN_CHANS: 3
+    MLP_RATIO: 0.0
+    PATCH_NORM: true
+    PATCH_SIZE: 4
+    SHARED_SSM: false
+    SOFTMAX: false
+    SSM_RATIO: 2.0
+OUTPUT: ./res_vmamba_cnf241_result_best/vssm_small/default
+PRINT_FREQ: 10
+SAVE_FREQ: 1
+SEED: 0
+TAG: default
+TEST:
+  CROP: true
+  SEQUENTIAL: false
+  SHUFFLE: false
+THROUGHPUT_MODE: false
+TRAIN:
+  ACCUMULATION_STEPS: 1
+  AUTO_RESUME: true
+  BASE_LR: 0.000125
+  CLIP_GRAD: 5.0
+  EPOCHS: 300
+  LAYER_DECAY: 1.0
+  LR_SCHEDULER:
+    DECAY_EPOCHS: 30
+    DECAY_RATE: 0.1
+    GAMMA: 0.1
+    MULTISTEPS: []
+    NAME: cosine
+    WARMUP_PREFIX: true
+  MIN_LR: 1.25e-06
+  MOE:
+    SAVE_MASTER: false
+  OPTIMIZER:
+    BETAS:
+    - 0.9
+    - 0.999
+    EPS: 1.0e-08
+    MOMENTUM: 0.9
+    NAME: adamw
+  START_EPOCH: 0
+  USE_CHECKPOINT: false
+  WARMUP_EPOCHS: 20
+  WARMUP_LR: 1.25e-07
+  WEIGHT_DECAY: 0.05
+
+[2024-02-22 18:03:56 vssm_small] (main.py 405): INFO {"cfg": "configs/vssm/vssm_small_224.yaml", "opts": null, "batch_size": 128, "data_path": "/home/public_3T/food_data/CNFOOD-241", "zip": false, "cache_mode": "part", "pretrained": "./res_vmamba_cnf241_result_2/vssm_small/default/ckpt_epoch_12.pth", "resume": null, "accumulation_steps": null, "use_checkpoint": false, "disable_amp": false, "amp_opt_level": null, "output": "./res_vmamba_cnf241_result_best", "tag": null, "eval": false, "throughput": false, "local_rank": 0, "fused_layernorm": false, "optim": null, "model_ema": true, "model_ema_decay": 0.9999, "model_ema_force_cpu": false}
+[2024-02-22 18:03:56 vssm_small] (main.py 112): INFO Creating model:vssm/vssm_small
+[2024-02-22 18:03:57 vssm_small] (main.py 118): INFO VSSM(
+  (patch_embed): Sequential(
+    (0): Conv2d(3, 96, kernel_size=(4, 4), stride=(4, 4))
+    (1): Permute()
+    (2): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
+  )
+  (layers): ModuleList(
+    (0): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=96, out_features=384, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(192, 192, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=192)
+            (out_proj): Linear(in_features=192, out_features=96, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.0)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=96, out_features=384, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(192, 192, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=192)
+            (out_proj): Linear(in_features=192, out_features=96, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.00937500037252903)
+        )
+      )
+      (downsample): PatchMerging2D(
+        (reduction): Linear(in_features=384, out_features=192, bias=False)
+        (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+      )
+    )
+    (1): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=192, out_features=768, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384)
+            (out_proj): Linear(in_features=384, out_features=192, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.01875000074505806)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=192, out_features=768, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(384, 384, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=384)
+            (out_proj): Linear(in_features=384, out_features=192, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.02812500111758709)
+        )
+      )
+      (downsample): PatchMerging2D(
+        (reduction): Linear(in_features=768, out_features=384, bias=False)
+        (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+      )
+    )
+    (2): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.03750000149011612)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.046875)
+        )
+        (2): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.05625000223517418)
+        )
+        (3): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.06562500447034836)
+        )
+        (4): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.07500000298023224)
+        )
+        (5): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.08437500149011612)
+        )
+        (6): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.09375)
+        )
+        (7): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.10312500596046448)
+        )
+        (8): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.11250000447034836)
+        )
+        (9): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.12187500298023224)
+        )
+        (10): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.13125000894069672)
+        )
+        (11): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.140625)
+        )
+        (12): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.15000000596046448)
+        )
+        (13): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.15937501192092896)
+        )
+        (14): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.16875000298023224)
+        )
+        (15): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.17812500894069672)
+        )
+        (16): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.1875)
+        )
+        (17): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.19687500596046448)
+        )
+        (18): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.20625001192092896)
+        )
+        (19): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.21562501788139343)
+        )
+        (20): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.22500000894069672)
+        )
+        (21): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.2343750149011612)
+        )
+        (22): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.24375000596046448)
+        )
+        (23): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.25312501192092896)
+        )
+        (24): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.26250001788139343)
+        )
+        (25): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.2718750238418579)
+        )
+        (26): VSSBlock(
+          (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=384, out_features=1536, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(768, 768, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=768)
+            (out_proj): Linear(in_features=768, out_features=384, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.28125)
+        )
+      )
+      (downsample): PatchMerging2D(
+        (reduction): Linear(in_features=1536, out_features=768, bias=False)
+        (norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
+      )
+    )
+    (3): Sequential(
+      (blocks): Sequential(
+        (0): VSSBlock(
+          (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=768, out_features=3072, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(1536, 1536, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1536)
+            (out_proj): Linear(in_features=1536, out_features=768, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.2906250059604645)
+        )
+        (1): VSSBlock(
+          (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+          (op): SS2D(
+            (out_norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
+            (in_proj): Linear(in_features=768, out_features=3072, bias=False)
+            (act): SiLU()
+            (conv2d): Conv2d(1536, 1536, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1536)
+            (out_proj): Linear(in_features=1536, out_features=768, bias=False)
+            (dropout): Identity()
+          )
+          (drop_path): timm.DropPath(0.30000001192092896)
+        )
+      )
+      (downsample): Identity()
+    )
+  )
+  (classifier): Sequential(
+    (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+    (permute): Permute()
+    (avgpool): AdaptiveAvgPool2d(output_size=1)
+    (flatten): Flatten(start_dim=1, end_dim=-1)
+    (head): Linear(in_features=768, out_features=1000, bias=True)
+  )
+)
+[2024-02-22 18:03:57 vssm_small] (main.py 120): INFO number of params: 44417416
+[2024-02-22 18:03:58 vssm_small] (main.py 123): INFO number of GFLOPs: 11.231522784
+[2024-02-22 18:03:58 vssm_small] (main.py 167): INFO auto resuming from ./res_vmamba_cnf241_result_best/vssm_small/default/ckpt_epoch_166.pth
+[2024-02-22 18:03:58 vssm_small] (utils.py 18): INFO ==============> Resuming form ./res_vmamba_cnf241_result_best/vssm_small/default/ckpt_epoch_166.pth....................
+[2024-02-22 18:04:00 vssm_small] (utils.py 27): INFO resuming model: <All keys matched successfully>
+[2024-02-22 18:04:00 vssm_small] (utils.py 34): INFO resuming model_ema: <All keys matched successfully>
+[2024-02-22 18:04:00 vssm_small] (utils.py 48): INFO => loaded successfully './res_vmamba_cnf241_result_best/vssm_small/default/ckpt_epoch_166.pth' (epoch 166)
+[2024-02-22 18:04:11 vssm_small] (main.py 324): INFO Test: [0/164]	Time 10.625 (10.625)	Loss 0.3557 (0.3557)	Acc@1 92.188 (92.188)	Acc@5 99.219 (99.219)	Mem 7155MB
+[2024-02-22 18:04:16 vssm_small] (main.py 324): INFO Test: [10/164]	Time 0.483 (1.405)	Loss 0.9014 (0.9043)	Acc@1 75.000 (78.622)	Acc@5 96.094 (95.099)	Mem 7155MB
+[2024-02-22 18:04:21 vssm_small] (main.py 324): INFO Test: [20/164]	Time 0.483 (0.966)	Loss 1.3975 (1.0178)	Acc@1 73.438 (75.818)	Acc@5 89.844 (95.238)	Mem 7155MB
+[2024-02-22 18:04:25 vssm_small] (main.py 324): INFO Test: [30/164]	Time 0.483 (0.810)	Loss 0.5830 (0.9851)	Acc@1 86.719 (77.344)	Acc@5 96.875 (94.960)	Mem 7155MB
+[2024-02-22 18:04:30 vssm_small] (main.py 324): INFO Test: [40/164]	Time 0.483 (0.730)	Loss 0.9902 (0.9247)	Acc@1 77.344 (78.925)	Acc@5 95.312 (95.332)	Mem 7155MB
+[2024-02-22 18:04:35 vssm_small] (main.py 324): INFO Test: [50/164]	Time 0.483 (0.682)	Loss 1.3584 (0.9845)	Acc@1 69.531 (77.681)	Acc@5 92.969 (94.838)	Mem 7155MB
+[2024-02-22 18:04:40 vssm_small] (main.py 324): INFO Test: [60/164]	Time 0.483 (0.649)	Loss 1.1582 (1.0527)	Acc@1 70.312 (75.999)	Acc@5 96.875 (94.237)	Mem 7155MB
+[2024-02-22 18:04:45 vssm_small] (main.py 324): INFO Test: [70/164]	Time 0.483 (0.626)	Loss 0.4968 (1.0371)	Acc@1 89.844 (76.320)	Acc@5 96.875 (94.311)	Mem 7155MB
+[2024-02-22 18:04:50 vssm_small] (main.py 324): INFO Test: [80/164]	Time 0.483 (0.608)	Loss 0.4280 (1.0560)	Acc@1 91.406 (75.965)	Acc@5 98.438 (94.088)	Mem 7155MB
+[2024-02-22 18:04:54 vssm_small] (main.py 324): INFO Test: [90/164]	Time 0.483 (0.594)	Loss 1.0479 (1.0186)	Acc@1 71.875 (76.829)	Acc@5 99.219 (94.420)	Mem 7155MB
+[2024-02-22 18:04:59 vssm_small] (main.py 324): INFO Test: [100/164]	Time 0.483 (0.583)	Loss 0.5444 (1.0171)	Acc@1 82.812 (77.158)	Acc@5 100.000 (94.307)	Mem 7155MB
+[2024-02-22 18:05:04 vssm_small] (main.py 324): INFO Test: [110/164]	Time 0.483 (0.574)	Loss 1.3740 (1.0362)	Acc@1 67.188 (76.464)	Acc@5 96.875 (94.348)	Mem 7155MB
+[2024-02-22 18:05:09 vssm_small] (main.py 324): INFO Test: [120/164]	Time 0.483 (0.567)	Loss 2.1602 (1.0386)	Acc@1 33.594 (76.220)	Acc@5 89.844 (94.441)	Mem 7155MB
+[2024-02-22 18:05:14 vssm_small] (main.py 324): INFO Test: [130/164]	Time 0.483 (0.560)	Loss 1.2930 (1.0532)	Acc@1 57.812 (75.889)	Acc@5 98.438 (94.281)	Mem 7155MB
+[2024-02-22 18:05:19 vssm_small] (main.py 324): INFO Test: [140/164]	Time 0.483 (0.555)	Loss 0.7490 (1.0376)	Acc@1 81.250 (76.141)	Acc@5 96.094 (94.437)	Mem 7155MB
+[2024-02-22 18:05:23 vssm_small] (main.py 324): INFO Test: [150/164]	Time 0.482 (0.550)	Loss 1.1650 (1.0309)	Acc@1 69.531 (76.293)	Acc@5 98.438 (94.521)	Mem 7155MB
+[2024-02-22 18:05:28 vssm_small] (main.py 324): INFO Test: [160/164]	Time 0.483 (0.546)	Loss 0.5903 (1.0296)	Acc@1 89.844 (76.305)	Acc@5 95.312 (94.580)	Mem 7155MB
+[2024-02-22 18:05:31 vssm_small] (main.py 331): INFO  * Acc@1 76.541 Acc@5 94.638
+[2024-02-22 18:05:31 vssm_small] (main.py 174): INFO Accuracy of the network on the 20943 test images: 76.5%
+[2024-02-22 18:05:39 vssm_small] (main.py 324): INFO Test: [0/164]	Time 8.835 (8.835)	Loss 0.4526 (0.4526)	Acc@1 89.844 (89.844)	Acc@5 99.219 (99.219)	Mem 7155MB
+[2024-02-22 18:05:44 vssm_small] (main.py 324): INFO Test: [10/164]	Time 0.482 (1.242)	Loss 1.1172 (0.8497)	Acc@1 67.969 (79.830)	Acc@5 96.094 (95.739)	Mem 7155MB
+[2024-02-22 18:05:49 vssm_small] (main.py 324): INFO Test: [20/164]	Time 0.483 (0.880)	Loss 1.3506 (0.9275)	Acc@1 72.656 (77.567)	Acc@5 92.188 (96.168)	Mem 7155MB
+[2024-02-22 18:05:54 vssm_small] (main.py 324): INFO Test: [30/164]	Time 0.483 (0.752)	Loss 0.6631 (0.9005)	Acc@1 84.375 (79.133)	Acc@5 96.875 (95.640)	Mem 7155MB
+[2024-02-22 18:05:59 vssm_small] (main.py 324): INFO Test: [40/164]	Time 0.483 (0.686)	Loss 0.8730 (0.8447)	Acc@1 78.906 (80.640)	Acc@5 96.094 (95.941)	Mem 7155MB
+[2024-02-22 18:06:03 vssm_small] (main.py 324): INFO Test: [50/164]	Time 0.483 (0.646)	Loss 1.4102 (0.9097)	Acc@1 66.406 (79.350)	Acc@5 92.969 (95.343)	Mem 7155MB
+[2024-02-22 18:06:08 vssm_small] (main.py 324): INFO Test: [60/164]	Time 0.482 (0.620)	Loss 1.1191 (0.9768)	Acc@1 67.969 (77.818)	Acc@5 96.875 (94.762)	Mem 7155MB
+[2024-02-22 18:06:13 vssm_small] (main.py 324): INFO Test: [70/164]	Time 0.482 (0.600)	Loss 0.4170 (0.9548)	Acc@1 90.625 (78.191)	Acc@5 96.094 (94.971)	Mem 7155MB
+[2024-02-22 18:06:18 vssm_small] (main.py 324): INFO Test: [80/164]	Time 0.482 (0.586)	Loss 0.4082 (0.9766)	Acc@1 91.406 (77.778)	Acc@5 98.438 (94.676)	Mem 7155MB
+[2024-02-22 18:06:23 vssm_small] (main.py 324): INFO Test: [90/164]	Time 0.482 (0.574)	Loss 1.0576 (0.9459)	Acc@1 72.656 (78.546)	Acc@5 99.219 (94.943)	Mem 7155MB
+[2024-02-22 18:06:28 vssm_small] (main.py 324): INFO Test: [100/164]	Time 0.483 (0.565)	Loss 0.5508 (0.9468)	Acc@1 84.375 (78.860)	Acc@5 100.000 (94.825)	Mem 7155MB
+[2024-02-22 18:06:32 vssm_small] (main.py 324): INFO Test: [110/164]	Time 0.483 (0.558)	Loss 1.1367 (0.9615)	Acc@1 68.750 (78.202)	Acc@5 97.656 (94.869)	Mem 7155MB
+[2024-02-22 18:06:37 vssm_small] (main.py 324): INFO Test: [120/164]	Time 0.482 (0.552)	Loss 2.1855 (0.9641)	Acc@1 29.688 (77.893)	Acc@5 91.406 (94.990)	Mem 7155MB
+[2024-02-22 18:06:42 vssm_small] (main.py 324): INFO Test: [130/164]	Time 0.483 (0.546)	Loss 1.2090 (0.9734)	Acc@1 60.156 (77.642)	Acc@5 99.219 (94.931)	Mem 7155MB
+[2024-02-22 18:06:47 vssm_small] (main.py 324): INFO Test: [140/164]	Time 0.483 (0.542)	Loss 0.6606 (0.9576)	Acc@1 82.812 (77.876)	Acc@5 99.219 (95.107)	Mem 7155MB
+[2024-02-22 18:06:52 vssm_small] (main.py 324): INFO Test: [150/164]	Time 0.482 (0.538)	Loss 0.9053 (0.9501)	Acc@1 77.344 (78.084)	Acc@5 98.438 (95.183)	Mem 7155MB
+[2024-02-22 18:06:57 vssm_small] (main.py 324): INFO Test: [160/164]	Time 0.482 (0.534)	Loss 0.5884 (0.9481)	Acc@1 86.719 (78.023)	Acc@5 96.875 (95.259)	Mem 7155MB
+[2024-02-22 18:06:58 vssm_small] (main.py 331): INFO  * Acc@1 78.260 Acc@5 95.306
+[2024-02-22 18:06:58 vssm_small] (main.py 177): INFO Accuracy of the network ema on the 20943 test images: 78.3%
+[2024-02-22 18:06:58 vssm_small] (main.py 196): INFO Start training
+[2024-02-22 18:07:10 vssm_small] (main.py 274): INFO Train: [167/300][0/933]	eta 3:02:27 lr 0.000058	 wd 0.0500	time 11.7339 (11.7339)	data time 8.6883 (8.6883)	loss 3.2837 (3.2837)	grad_norm 7.2753 (7.2753)	loss_scale 32768.0000 (32768.0000)	mem 50097MB
+[2024-02-22 18:07:26 vssm_small] (main.py 274): INFO Train: [167/300][10/933]	eta 0:38:21 lr 0.000058	 wd 0.0500	time 1.5683 (2.4935)	data time 0.0006 (0.7903)	loss 2.0028 (3.0696)	grad_norm 5.6539 (6.8326)	loss_scale 32768.0000 (32768.0000)	mem 50285MB
+[2024-02-22 18:07:41 vssm_small] (main.py 274): INFO Train: [167/300][20/933]	eta 0:31:15 lr 0.000058	 wd 0.0500	time 1.5680 (2.0546)	data time 0.0006 (0.4143)	loss 2.7450 (2.9151)	grad_norm 4.8270 (6.2856)	loss_scale 32768.0000 (32768.0000)	mem 50285MB