Training from n epoch
manhcntt21 opened this issue · 0 comments
manhcntt21 commented
Hi
I have a question. I trained with my custom dataset, assum 2nd epoch. Now, i want to continue training from this 2nd epoch. Is there any way to do something like that?
I replaced TRAIN.CHECKPOINT_FILE_PATH with value is 2nd epoch's checkpoint, but maybe not working.
This is log
$ bash ./exp/uniformer_s8x8_k400/run.sh
...
[11/17 12:04:36][INFO] uniformer.py: 287: Use checkpoint: True
[11/17 12:04:36][INFO] uniformer.py: 288: Checkpoint number: [0, 0, 4, 0]
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: patch_embed1.proj.weight, torch.Size([64, 3, 4, 4]) => torch.Size([64, 3, 3, 4, 4])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: patch_embed2.proj.weight, torch.Size([128, 64, 2, 2]) => torch.Size([128, 64, 1, 2, 2])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: patch_embed3.proj.weight, torch.Size([320, 128, 2, 2]) => torch.Size([320, 128, 1, 2, 2])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: patch_embed4.proj.weight, torch.Size([512, 320, 2, 2]) => torch.Size([512, 320, 1, 2, 2])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.0.pos_embed.weight, torch.Size([64, 1, 3, 3]) => torch.Size([64, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.0.conv1.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.0.conv2.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.0.attn.weight, torch.Size([64, 1, 5, 5]) => torch.Size([64, 1, 5, 5, 5])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.0.mlp.fc1.weight, torch.Size([256, 64, 1, 1]) => torch.Size([256, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.0.mlp.fc2.weight, torch.Size([64, 256, 1, 1]) => torch.Size([64, 256, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.1.pos_embed.weight, torch.Size([64, 1, 3, 3]) => torch.Size([64, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.1.conv1.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.1.conv2.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.1.attn.weight, torch.Size([64, 1, 5, 5]) => torch.Size([64, 1, 5, 5, 5])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.1.mlp.fc1.weight, torch.Size([256, 64, 1, 1]) => torch.Size([256, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.1.mlp.fc2.weight, torch.Size([64, 256, 1, 1]) => torch.Size([64, 256, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.2.pos_embed.weight, torch.Size([64, 1, 3, 3]) => torch.Size([64, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.2.conv1.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.2.conv2.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.2.attn.weight, torch.Size([64, 1, 5, 5]) => torch.Size([64, 1, 5, 5, 5])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.2.mlp.fc1.weight, torch.Size([256, 64, 1, 1]) => torch.Size([256, 64, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks1.2.mlp.fc2.weight, torch.Size([64, 256, 1, 1]) => torch.Size([64, 256, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.0.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.0.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.0.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.0.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.0.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.0.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.1.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.1.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.1.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.1.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.1.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.1.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.2.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.2.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.2.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.2.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.2.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.2.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.3.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.3.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.3.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.3.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.3.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks2.3.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.0.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.1.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.2.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.3.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.4.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.5.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.6.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks3.7.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks4.0.pos_embed.weight, torch.Size([512, 1, 3, 3]) => torch.Size([512, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks4.1.pos_embed.weight, torch.Size([512, 1, 3, 3]) => torch.Size([512, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 412: Inflate: blocks4.2.pos_embed.weight, torch.Size([512, 1, 3, 3]) => torch.Size([512, 1, 3, 3, 3])
[11/17 12:04:36][INFO] uniformer.py: 410: Ignore: head.weight
[11/17 12:04:36][INFO] uniformer.py: 410: Ignore: head.bias
[11/17 12:04:36][INFO] build.py: 45: load pretrained model
[11/17 12:04:37][INFO] misc.py: 183: Model:
...
[11/17 12:04:37][INFO] misc.py: 184: Params: 21,400,400
[11/17 12:04:37][INFO] misc.py: 185: Mem: 0.0800790786743164 MB
e:\master\uniformer\video_classification\slowfast\models\uniformer.py:85: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
[11/17 12:04:39][WARNING] jit_analysis.py: 499: Unsupported operator aten::add encountered 42 time(s)
[11/17 12:04:39][WARNING] jit_analysis.py: 499: Unsupported operator aten::gelu encountered 14 time(s)
[11/17 12:04:39][WARNING] jit_analysis.py: 499: Unsupported operator prim::PythonOp.CheckpointFunction encountered 4 time(s)
[11/17 12:04:39][WARNING] jit_analysis.py: 499: Unsupported operator aten::div encountered 7 time(s)
[11/17 12:04:39][WARNING] jit_analysis.py: 499: Unsupported operator aten::mul encountered 7 time(s)
[11/17 12:04:39][WARNING] jit_analysis.py: 499: Unsupported operator aten::softmax encountered 7 time(s)
[11/17 12:04:39][WARNING] jit_analysis.py: 499: Unsupported operator aten::mean encountered 1 time(s)
[11/17 12:04:39][WARNING] jit_analysis.py: 511: The following submodules of the model were never called during the trace of the graph. They may be unused, or they were accessed by direct calls to .forward() or via other python methods. In the latter case they will have zeros for statistics, though their statistics will still contribute to their parent calling module.
blocks1.1.drop_path, blocks1.2.drop_path, blocks2.0.drop_path, blocks2.1.drop_path, blocks2.2.drop_path, blocks2.3.drop_path, blocks3.0, blocks3.0.attn, blocks3.0.attn.attn_drop, blocks3.0.attn.proj, blocks3.0.attn.proj_drop, blocks3.0.attn.qkv, blocks3.0.drop_path, blocks3.0.mlp, blocks3.0.mlp.act, blocks3.0.mlp.drop, blocks3.0.mlp.fc1, blocks3.0.mlp.fc2, blocks3.0.norm1, blocks3.0.norm2, blocks3.0.pos_embed, blocks3.1, blocks3.1.attn, blocks3.1.attn.attn_drop, blocks3.1.attn.proj, blocks3.1.attn.proj_drop, blocks3.1.attn.qkv, blocks3.1.drop_path, blocks3.1.mlp, blocks3.1.mlp.act, blocks3.1.mlp.drop, blocks3.1.mlp.fc1, blocks3.1.mlp.fc2, blocks3.1.norm1, blocks3.1.norm2, blocks3.1.pos_embed, blocks3.2, blocks3.2.attn, blocks3.2.attn.attn_drop, blocks3.2.attn.proj, blocks3.2.attn.proj_drop, blocks3.2.attn.qkv, blocks3.2.drop_path, blocks3.2.mlp, blocks3.2.mlp.act, blocks3.2.mlp.drop, blocks3.2.mlp.fc1, blocks3.2.mlp.fc2, blocks3.2.norm1, blocks3.2.norm2, blocks3.2.pos_embed, blocks3.3, blocks3.3.attn, blocks3.3.attn.attn_drop, blocks3.3.attn.proj, blocks3.3.attn.proj_drop, blocks3.3.attn.qkv, blocks3.3.drop_path, blocks3.3.mlp, blocks3.3.mlp.act, blocks3.3.mlp.drop, blocks3.3.mlp.fc1, blocks3.3.mlp.fc2, blocks3.3.norm1, blocks3.3.norm2, blocks3.3.pos_embed, blocks3.4.drop_path, blocks3.5.drop_path, blocks3.6.drop_path, blocks3.7.drop_path, blocks4.0.drop_path, blocks4.1.drop_path, blocks4.2.drop_path
[11/17 12:04:39][INFO] misc.py: 186: Flops: 12.149269504 G
e:\master\uniformer\video_classification\slowfast\models\uniformer.py:85: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::layer_norm encountered 18 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::add encountered 42 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::batch_norm encountered 15 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::gelu encountered 14 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator prim::PythonOp.CheckpointFunction encountered 4 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::div encountered 7 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::mul encountered 7 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::softmax encountered 7 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 499: Unsupported operator aten::mean encountered 1 time(s)
[11/17 12:04:40][WARNING] jit_analysis.py: 511: The following submodules of the model were never called during the trace of the graph. They may be unused, or they were accessed by direct calls to .forward() or via other python methods. In the latter case they will have zeros for statistics, though their statistics will still contribute to their parent calling module.
blocks1.1.drop_path, blocks1.2.drop_path, blocks2.0.drop_path, blocks2.1.drop_path, blocks2.2.drop_path, blocks2.3.drop_path, blocks3.0, blocks3.0.attn, blocks3.0.attn.attn_drop, blocks3.0.attn.proj, blocks3.0.attn.proj_drop, blocks3.0.attn.qkv, blocks3.0.drop_path, blocks3.0.mlp, blocks3.0.mlp.act, blocks3.0.mlp.drop, blocks3.0.mlp.fc1, blocks3.0.mlp.fc2, blocks3.0.norm1, blocks3.0.norm2, blocks3.0.pos_embed, blocks3.1, blocks3.1.attn, blocks3.1.attn.attn_drop, blocks3.1.attn.proj, blocks3.1.attn.proj_drop, blocks3.1.attn.qkv, blocks3.1.drop_path, blocks3.1.mlp, blocks3.1.mlp.act, blocks3.1.mlp.drop, blocks3.1.mlp.fc1, blocks3.1.mlp.fc2, blocks3.1.norm1, blocks3.1.norm2, blocks3.1.pos_embed, blocks3.2, blocks3.2.attn, blocks3.2.attn.attn_drop, blocks3.2.attn.proj, blocks3.2.attn.proj_drop, blocks3.2.attn.qkv, blocks3.2.drop_path, blocks3.2.mlp, blocks3.2.mlp.act, blocks3.2.mlp.drop, blocks3.2.mlp.fc1, blocks3.2.mlp.fc2, blocks3.2.norm1, blocks3.2.norm2, blocks3.2.pos_embed, blocks3.3, blocks3.3.attn, blocks3.3.attn.attn_drop, blocks3.3.attn.proj, blocks3.3.attn.proj_drop, blocks3.3.attn.qkv, blocks3.3.drop_path, blocks3.3.mlp, blocks3.3.mlp.act, blocks3.3.mlp.drop, blocks3.3.mlp.fc1, blocks3.3.mlp.fc2, blocks3.3.norm1, blocks3.3.norm2, blocks3.3.pos_embed, blocks3.4.drop_path, blocks3.5.drop_path, blocks3.6.drop_path, blocks3.7.drop_path, blocks4.0.drop_path, blocks4.1.drop_path, blocks4.2.drop_path
[11/17 12:04:40][INFO] misc.py: 191: Activations: 65.24801599999999 M
[11/17 12:04:40][INFO] misc.py: 196: nvidia-smi
Thu Nov 17 12:04:40 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 522.25 Driver Version: 522.25 CUDA Version: 11.8 |
|-------------------------------+----------------------+----------------------+
| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce ... WDDM | 00000000:26:00.0 On | N/A |
| 0% 51C P2 28W / 120W | 2424MiB / 6144MiB | 11% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 1456 C+G ...\PowerToys.FancyZones.exe N/A |
| 0 N/A N/A 5844 C+G ...werToys.PowerLauncher.exe N/A |
| 0 N/A N/A 9248 C+G ...6.0.3.0\GoogleDriveFS.exe N/A |
| 0 N/A N/A 10196 C+G C:\Windows\explorer.exe N/A |
| 0 N/A N/A 11424 C+G ...ropbox\Client\Dropbox.exe N/A |
| 0 N/A N/A 11960 C+G ...8bbwe\Microsoft.Notes.exe N/A |
| 0 N/A N/A 12932 C+G ...5n1h2txyewy\SearchApp.exe N/A |
| 0 N/A N/A 13532 C+G ...bbwe\Microsoft.Photos.exe N/A |
| 0 N/A N/A 13536 C+G ...me\Application\chrome.exe N/A |
| 0 N/A N/A 15480 C+G ...2txyewy\TextInputHost.exe N/A |
| 0 N/A N/A 16648 C+G ...ram Files\LGHUB\lghub.exe N/A |
| 0 N/A N/A 17376 C+G ...perience\NVIDIA Share.exe N/A |
| 0 N/A N/A 19824 C ...envs\uniformer\python.exe N/A |
| 0 N/A N/A 20416 C+G ...ons\Grammarly.Desktop.exe N/A |
| 0 N/A N/A 22376 C+G ...werToys.ColorPickerUI.exe N/A |
| 0 N/A N/A 24568 C+G ...y\ShellExperienceHost.exe N/A |
| 0 N/A N/A 26792 C+G ...zpdnekdrzrea0\Spotify.exe N/A |
| 0 N/A N/A 28376 C+G ...icrosoft VS Code\Code.exe N/A |
+-----------------------------------------------------------------------------+
bn 30, non bn 102, zero 154
[11/17 12:04:40][INFO] checkpoint_amp.py: 501: Load from last checkpoint, ./exp/uniformer_s8x8_k400\checkpoints\checkpoint_epoch_00002.pyth.
[11/17 12:04:40][INFO] checkpoint_amp.py: 213: Loading network weights from ./exp/uniformer_s8x8_k400\checkpoints\checkpoint_epoch_00002.pyth.
[11/17 12:04:40][INFO] kinetics.py: 76: Constructing Kinetics train...
[11/17 12:04:40][INFO] kinetics.py: 123: Constructing kinetics dataloader (size: 1275) from ./data_vid\train.csv
[11/17 12:04:40][INFO] kinetics.py: 76: Constructing Kinetics val...
[11/17 12:04:40][INFO] kinetics.py: 123: Constructing kinetics dataloader (size: 200) from ./data_vid\val.csv
[11/17 12:04:40][INFO] tensorboard_vis.py: 54: To see logged results in Tensorboard, please launch using the command `tensorboard --port=<port-number> --logdir ./exp/uniformer_s8x8_k400\runs-kinetics`
[11/17 12:04:40][INFO] train_net.py: 451: Start epoch: 3
D:\CAIDATPHANMEM\miniconda3\envs\uniformer\lib\site-packages\torchvision\transforms\_functional_video.py:5: UserWarning: The _functional_video module is deprecated. Please use the functional module instead.
warnings.warn(
D:\CAIDATPHANMEM\miniconda3\envs\uniformer\lib\site-packages\torchvision\transforms\_transforms_video.py:25: UserWarning: The _transforms_video module is deprecated. Please use the transforms module instead.
warnings.warn(
[11/17 12:04:43][INFO] test_net.py: 157: Test with config:
[11/17 12:04:43][INFO] test_net.py: 158: AUG:
AA_TYPE: rand-m7-n4-mstd0.5-inc1
COLOR_JITTER: 0.4
ENABLE: True
INTERPOLATION: bicubic
NUM_SAMPLE: 2
RE_COUNT: 1
RE_MODE: pixel
RE_PROB: 0.25
RE_SPLIT: False
AVA:
ANNOTATION_DIR: /mnt/vol/gfsai-flash3-east/ai-group/users/haoqifan/ava/frame_list/
BGR: False
DETECTION_SCORE_THRESH: 0.9
EXCLUSION_FILE: ava_val_excluded_timestamps_v2.2.csv
FRAME_DIR: /mnt/fair-flash3-east/ava_trainval_frames.img/
FRAME_LIST_DIR: /mnt/vol/gfsai-flash3-east/ai-group/users/haoqifan/ava/frame_list/
FULL_TEST_ON_VAL: False
GROUNDTRUTH_FILE: ava_val_v2.2.csv
IMG_PROC_BACKEND: cv2
LABEL_MAP_FILE: ava_action_list_v2.2_for_activitynet_2019.pbtxt
TEST_FORCE_FLIP: False
TEST_LISTS: ['val.csv']
TEST_PREDICT_BOX_LISTS: ['ava_val_predicted_boxes.csv']
TRAIN_GT_BOX_LISTS: ['ava_train_v2.2.csv']
TRAIN_LISTS: ['train.csv']
TRAIN_PCA_JITTER_ONLY: True
TRAIN_PREDICT_BOX_LISTS: []
TRAIN_USE_COLOR_AUGMENTATION: False
BENCHMARK:
LOG_PERIOD: 100
NUM_EPOCHS: 5
SHUFFLE: True
BN:
NORM_TYPE: batchnorm
NUM_BATCHES_PRECISE: 200
NUM_SPLITS: 1
NUM_SYNC_DEVICES: 1
USE_PRECISE_STATS: False
WEIGHT_DECAY: 0.0
DATA:
DECODING_BACKEND: decord
ENSEMBLE_METHOD: sum
IMAGE_TEMPLATE: {:05d}.jpg
INPUT_CHANNEL_NUM: [3]
INV_UNIFORM_SAMPLE: False
LABEL_PATH_TEMPLATE: somesomev1_rgb_{}_split.txt
MEAN: [0.45, 0.45, 0.45]
MULTI_LABEL: False
NUM_FRAMES: 8
PATH_LABEL_SEPARATOR: ,
PATH_PREFIX:
PATH_TO_DATA_DIR: ./data_vid
PATH_TO_PRELOAD_IMDB:
RANDOM_FLIP: True
REVERSE_INPUT_CHANNEL: False
SAMPLING_RATE: 8
STD: [0.225, 0.225, 0.225]
TARGET_FPS: 30
TEST_CROP_SIZE: 224
TRAIN_CROP_SIZE: 224
TRAIN_JITTER_ASPECT_RELATIVE: [0.75, 1.3333]
TRAIN_JITTER_MOTION_SHIFT: False
TRAIN_JITTER_SCALES: [256, 320]
TRAIN_JITTER_SCALES_RELATIVE: [0.08, 1.0]
TRAIN_PCA_EIGVAL: [0.225, 0.224, 0.229]
TRAIN_PCA_EIGVEC: [[-0.5675, 0.7192, 0.4009], [-0.5808, -0.0045, -0.814], [-0.5836, -0.6948, 0.4203]]
USE_OFFSET_SAMPLING: True
DATA_LOADER:
ENABLE_MULTI_THREAD_DECODE: False
NUM_WORKERS: 8
PIN_MEMORY: True
DEMO:
BUFFER_SIZE: 0
CLIP_VIS_SIZE: 10
COMMON_CLASS_NAMES: ['watch (a person)', 'talk to (e.g., self, a person, a group)', 'listen to (a person)', 'touch (an object)', 'carry/hold (an object)', 'walk', 'sit', 'lie/sleep', 'bend/bow (at the waist)']
COMMON_CLASS_THRES: 0.7
DETECTRON2_CFG: COCO-Detection/faster_rcnn_R_50_FPN_3x.yaml
DETECTRON2_THRESH: 0.9
DETECTRON2_WEIGHTS: detectron2://COCO-Detection/faster_rcnn_R_50_FPN_3x/137849458/model_final_280758.pkl
DISPLAY_HEIGHT: 0
DISPLAY_WIDTH: 0
ENABLE: False
FPS: 30
GT_BOXES:
INPUT_FORMAT: BGR
INPUT_VIDEO:
LABEL_FILE_PATH:
NUM_CLIPS_SKIP: 0
NUM_VIS_INSTANCES: 2
OUTPUT_FILE:
OUTPUT_FPS: -1
PREDS_BOXES:
SLOWMO: 1
STARTING_SECOND: 900
THREAD_ENABLE: False
UNCOMMON_CLASS_THRES: 0.3
VIS_MODE: thres
WEBCAM: -1
DETECTION:
ALIGNED: True
ENABLE: False
ROI_XFORM_RESOLUTION: 7
SPATIAL_SCALE_FACTOR: 16
DIST_BACKEND: gloo
LOG_MODEL_INFO: True
LOG_PERIOD: 10
MIXUP:
ALPHA: 0.8
CUTMIX_ALPHA: 1.0
ENABLE: True
LABEL_SMOOTH_VALUE: 0.1
PROB: 1.0
SWITCH_PROB: 0.5
MODEL:
ARCH: uniformer
CHECKPOINT_NUM: [0, 0, 4, 0]
DROPCONNECT_RATE: 0.0
DROPOUT_RATE: 0.5
FC_INIT_STD: 0.01
HEAD_ACT: softmax
LOSS_FUNC: soft_cross_entropy
MODEL_NAME: Uniformer
MULTI_PATHWAY_ARCH: ['slowfast']
NUM_CLASSES: 400
SINGLE_PATHWAY_ARCH: ['2d', 'c2d', 'i3d', 'slow', 'x3d', 'mvit', 'uniformer']
USE_CHECKPOINT: True
MULTIGRID:
BN_BASE_SIZE: 8
DEFAULT_B: 0
DEFAULT_S: 0
DEFAULT_T: 0
EPOCH_FACTOR: 1.5
EVAL_FREQ: 3
LONG_CYCLE: False
LONG_CYCLE_FACTORS: [(0.25, 0.7071067811865476), (0.5, 0.7071067811865476), (0.5, 1), (1, 1)]
LONG_CYCLE_SAMPLING_RATE: 0
SHORT_CYCLE: False
SHORT_CYCLE_FACTORS: [0.5, 0.7071067811865476]
MVIT:
CLS_EMBED_ON: True
DEPTH: 16
DIM_MUL: []
DROPOUT_RATE: 0.0
DROPPATH_RATE: 0.1
EMBED_DIM: 96
HEAD_MUL: []
MLP_RATIO: 4.0
MODE: conv
NORM: layernorm
NORM_STEM: False
NUM_HEADS: 1
PATCH_2D: False
PATCH_KERNEL: [3, 7, 7]
PATCH_PADDING: [2, 4, 4]
PATCH_STRIDE: [2, 4, 4]
POOL_KVQ_KERNEL: None
POOL_KV_STRIDE: []
POOL_Q_STRIDE: []
QKV_BIAS: True
SEP_POS_EMBED: False
ZERO_DECAY_POS_CLS: True
NONLOCAL:
GROUP: [[1], [1], [1], [1]]
INSTANTIATION: dot_product
LOCATION: [[[]], [[]], [[]], [[]]]
POOL: [[[1, 2, 2], [1, 2, 2]], [[1, 2, 2], [1, 2, 2]], [[1, 2, 2], [1, 2, 2]], [[1, 2, 2], [1, 2, 2]]]
NUM_GPUS: 1
NUM_SHARDS: 1
OUTPUT_DIR: ./exp/uniformer_s8x8_k400
RESNET:
DEPTH: 50
INPLACE_RELU: True
NUM_BLOCK_TEMP_KERNEL: [[3], [4], [6], [3]]
NUM_GROUPS: 1
SPATIAL_DILATIONS: [[1], [1], [1], [1]]
SPATIAL_STRIDES: [[1], [2], [2], [2]]
STRIDE_1X1: False
TRANS_FUNC: bottleneck_transform
WIDTH_PER_GROUP: 64
ZERO_INIT_FINAL_BN: False
RNG_SEED: 6666
SHARD_ID: 0
SLOWFAST:
ALPHA: 8
BETA_INV: 8
FUSION_CONV_CHANNEL_RATIO: 2
FUSION_KERNEL_SZ: 5
SOLVER:
BASE_LR: 0.0004
BASE_LR_SCALE_NUM_SHARDS: True
CLIP_GRADIENT: 20
COSINE_AFTER_WARMUP: True
COSINE_END_LR: 1e-06
DAMPENING: 0.0
GAMMA: 0.1
LRS: []
LR_POLICY: cosine
MAX_EPOCH: 2
MOMENTUM: 0.9
NESTEROV: True
OPTIMIZING_METHOD: adamw
STEPS: []
STEP_SIZE: 1
WARMUP_EPOCHS: 10.0
WARMUP_FACTOR: 0.1
WARMUP_START_LR: 1e-06
WEIGHT_DECAY: 0.05
ZERO_WD_1D_PARAM: True
TENSORBOARD:
CATEGORIES_PATH:
CLASS_NAMES_PATH:
CONFUSION_MATRIX:
ENABLE: False
FIGSIZE: [8, 8]
SUBSET_PATH:
ENABLE: True
HISTOGRAM:
ENABLE: False
FIGSIZE: [8, 8]
SUBSET_PATH:
TOPK: 10
LOG_DIR:
MODEL_VIS:
ACTIVATIONS: False
COLORMAP: Pastel2
ENABLE: False
GRAD_CAM:
COLORMAP: viridis
ENABLE: True
LAYER_LIST: []
USE_TRUE_LABEL: False
INPUT_VIDEO: False
LAYER_LIST: []
MODEL_WEIGHTS: False
TOPK_PREDS: 1
PREDICTIONS_PATH:
WRONG_PRED_VIS:
ENABLE: False
SUBSET_PATH:
TAG: Incorrectly classified videos.
TEST:
BATCH_SIZE: 64
CHECKPOINT_FILE_PATH:
CHECKPOINT_TYPE: pytorch
DATASET: kinetics
ENABLE: True
NUM_ENSEMBLE_VIEWS: 1
NUM_SPATIAL_CROPS: 1
SAVE_RESULTS_PATH:
TRAIN:
AUTO_RESUME: True
BATCH_SIZE: 8
CHECKPOINT_CLEAR_NAME_PATTERN: ()
CHECKPOINT_EPOCH_RESET: False
CHECKPOINT_FILE_PATH: ./exp/uniformer_s8x8_k400/checkpoints/checkpoint_epoch_00002.pyth
CHECKPOINT_INFLATE: False
CHECKPOINT_PERIOD: 1
CHECKPOINT_TYPE: pytorch
DATASET: kinetics
ENABLE: True
EVAL_PERIOD: 5
UNIFORMER:
ATTENTION_DROPOUT_RATE: 0
DEPTH: [3, 4, 8, 3]
DROPOUT_RATE: 0
DROP_DEPTH_RATE: 0.1
EMBED_DIM: [64, 128, 320, 512]
HEAD_DIM: 64
MLP_RATIO: 4
PRETRAIN_NAME: uniformer_small_in1k
QKV_BIAS: True
QKV_SCALE: None
REPRESENTATION_SIZE: None
SPLIT: False
STAGE_TYPE: [0, 0, 1, 1]
STD: False
X3D:
BN_LIN5: False
BOTTLENECK_FACTOR: 1.0
CHANNELWISE_3x3x3: True
DEPTH_FACTOR: 1.0
DIM_C1: 12
DIM_C5: 2048
SCALE_RES2: False
WIDTH_FACTOR: 1.0
[11/17 12:04:43][INFO] uniformer.py: 287: Use checkpoint: True
[11/17 12:04:43][INFO] uniformer.py: 288: Checkpoint number: [0, 0, 4, 0]
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: patch_embed1.proj.weight, torch.Size([64, 3, 4, 4]) => torch.Size([64, 3, 3, 4, 4])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: patch_embed2.proj.weight, torch.Size([128, 64, 2, 2]) => torch.Size([128, 64, 1, 2, 2])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: patch_embed3.proj.weight, torch.Size([320, 128, 2, 2]) => torch.Size([320, 128, 1, 2, 2])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: patch_embed4.proj.weight, torch.Size([512, 320, 2, 2]) => torch.Size([512, 320, 1, 2, 2])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.0.pos_embed.weight, torch.Size([64, 1, 3, 3]) => torch.Size([64, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.0.conv1.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.0.conv2.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.0.attn.weight, torch.Size([64, 1, 5, 5]) => torch.Size([64, 1, 5, 5, 5])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.0.mlp.fc1.weight, torch.Size([256, 64, 1, 1]) => torch.Size([256, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.0.mlp.fc2.weight, torch.Size([64, 256, 1, 1]) => torch.Size([64, 256, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.1.pos_embed.weight, torch.Size([64, 1, 3, 3]) => torch.Size([64, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.1.conv1.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.1.conv2.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.1.attn.weight, torch.Size([64, 1, 5, 5]) => torch.Size([64, 1, 5, 5, 5])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.1.mlp.fc1.weight, torch.Size([256, 64, 1, 1]) => torch.Size([256, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.1.mlp.fc2.weight, torch.Size([64, 256, 1, 1]) => torch.Size([64, 256, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.2.pos_embed.weight, torch.Size([64, 1, 3, 3]) => torch.Size([64, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.2.conv1.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.2.conv2.weight, torch.Size([64, 64, 1, 1]) => torch.Size([64, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.2.attn.weight, torch.Size([64, 1, 5, 5]) => torch.Size([64, 1, 5, 5, 5])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.2.mlp.fc1.weight, torch.Size([256, 64, 1, 1]) => torch.Size([256, 64, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks1.2.mlp.fc2.weight, torch.Size([64, 256, 1, 1]) => torch.Size([64, 256, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.0.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.0.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.0.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.0.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.0.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.0.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.1.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.1.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.1.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.1.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.1.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.1.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.2.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.2.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.2.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.2.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.2.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.2.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.3.pos_embed.weight, torch.Size([128, 1, 3, 3]) => torch.Size([128, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.3.conv1.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.3.conv2.weight, torch.Size([128, 128, 1, 1]) => torch.Size([128, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.3.attn.weight, torch.Size([128, 1, 5, 5]) => torch.Size([128, 1, 5, 5, 5])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.3.mlp.fc1.weight, torch.Size([512, 128, 1, 1]) => torch.Size([512, 128, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks2.3.mlp.fc2.weight, torch.Size([128, 512, 1, 1]) => torch.Size([128, 512, 1, 1, 1])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.0.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.1.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.2.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.3.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.4.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.5.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.6.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks3.7.pos_embed.weight, torch.Size([320, 1, 3, 3]) => torch.Size([320, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks4.0.pos_embed.weight, torch.Size([512, 1, 3, 3]) => torch.Size([512, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks4.1.pos_embed.weight, torch.Size([512, 1, 3, 3]) => torch.Size([512, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 412: Inflate: blocks4.2.pos_embed.weight, torch.Size([512, 1, 3, 3]) => torch.Size([512, 1, 3, 3, 3])
[11/17 12:04:43][INFO] uniformer.py: 410: Ignore: head.weight
[11/17 12:04:43][INFO] uniformer.py: 410: Ignore: head.bias
[11/17 12:04:43][INFO] build.py: 45: load pretrained model
[11/17 12:04:43][INFO] misc.py: 183: Model:
...
...
...
[11/17 12:04:43][INFO] misc.py: 184: Params: 21,400,400
[11/17 12:04:43][INFO] misc.py: 185: Mem: 0.0800790786743164 MB
e:\master\uniformer\video_classification\slowfast\models\uniformer.py:85: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::add encountered 42 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::gelu encountered 14 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator prim::PythonOp.CheckpointFunction encountered 4 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::div encountered 7 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::mul encountered 7 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::softmax encountered 7 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::mean encountered 1 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 511: The following submodules of the model were never called during the trace of the graph. They may be unused, or they were accessed by direct calls to .forward() or via other python methods. In the latter case they will have zeros for statistics, though their statistics will still contribute to their parent calling module.
blocks1.1.drop_path, blocks1.2.drop_path, blocks2.0.drop_path, blocks2.1.drop_path, blocks2.2.drop_path, blocks2.3.drop_path, blocks3.0, blocks3.0.attn, blocks3.0.attn.attn_drop, blocks3.0.attn.proj, blocks3.0.attn.proj_drop, blocks3.0.attn.qkv, blocks3.0.drop_path, blocks3.0.mlp, blocks3.0.mlp.act, blocks3.0.mlp.drop, blocks3.0.mlp.fc1, blocks3.0.mlp.fc2, blocks3.0.norm1, blocks3.0.norm2, blocks3.0.pos_embed, blocks3.1, blocks3.1.attn, blocks3.1.attn.attn_drop, blocks3.1.attn.proj, blocks3.1.attn.proj_drop, blocks3.1.attn.qkv, blocks3.1.drop_path, blocks3.1.mlp, blocks3.1.mlp.act, blocks3.1.mlp.drop, blocks3.1.mlp.fc1, blocks3.1.mlp.fc2, blocks3.1.norm1, blocks3.1.norm2, blocks3.1.pos_embed, blocks3.2, blocks3.2.attn, blocks3.2.attn.attn_drop, blocks3.2.attn.proj, blocks3.2.attn.proj_drop, blocks3.2.attn.qkv, blocks3.2.drop_path, blocks3.2.mlp, blocks3.2.mlp.act, blocks3.2.mlp.drop, blocks3.2.mlp.fc1, blocks3.2.mlp.fc2, blocks3.2.norm1, blocks3.2.norm2, blocks3.2.pos_embed, blocks3.3, blocks3.3.attn, blocks3.3.attn.attn_drop, blocks3.3.attn.proj, blocks3.3.attn.proj_drop, blocks3.3.attn.qkv, blocks3.3.drop_path, blocks3.3.mlp, blocks3.3.mlp.act, blocks3.3.mlp.drop, blocks3.3.mlp.fc1, blocks3.3.mlp.fc2, blocks3.3.norm1, blocks3.3.norm2, blocks3.3.pos_embed, blocks3.4.drop_path, blocks3.5.drop_path, blocks3.6.drop_path, blocks3.7.drop_path, blocks4.0.drop_path, blocks4.1.drop_path, blocks4.2.drop_path
[11/17 12:04:46][INFO] misc.py: 186: Flops: 12.149269504 G
e:\master\uniformer\video_classification\slowfast\models\uniformer.py:85: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::layer_norm encountered 18 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::add encountered 42 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::batch_norm encountered 15 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::gelu encountered 14 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator prim::PythonOp.CheckpointFunction encountered 4 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::div encountered 7 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::mul encountered 7 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::softmax encountered 7 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 499: Unsupported operator aten::mean encountered 1 time(s)
[11/17 12:04:46][WARNING] jit_analysis.py: 511: The following submodules of the model were never called during the trace of the graph. They may be unused, or they were accessed by direct calls to .forward() or via other python methods. In the latter case they will have zeros for statistics, though their statistics will still contribute to their parent calling module.
blocks1.1.drop_path, blocks1.2.drop_path, blocks2.0.drop_path, blocks2.1.drop_path, blocks2.2.drop_path, blocks2.3.drop_path, blocks3.0, blocks3.0.attn, blocks3.0.attn.attn_drop, blocks3.0.attn.proj, blocks3.0.attn.proj_drop, blocks3.0.attn.qkv, blocks3.0.drop_path, blocks3.0.mlp, blocks3.0.mlp.act, blocks3.0.mlp.drop, blocks3.0.mlp.fc1, blocks3.0.mlp.fc2, blocks3.0.norm1, blocks3.0.norm2, blocks3.0.pos_embed, blocks3.1, blocks3.1.attn, blocks3.1.attn.attn_drop, blocks3.1.attn.proj, blocks3.1.attn.proj_drop, blocks3.1.attn.qkv, blocks3.1.drop_path, blocks3.1.mlp, blocks3.1.mlp.act, blocks3.1.mlp.drop, blocks3.1.mlp.fc1, blocks3.1.mlp.fc2, blocks3.1.norm1, blocks3.1.norm2, blocks3.1.pos_embed, blocks3.2, blocks3.2.attn, blocks3.2.attn.attn_drop, blocks3.2.attn.proj, blocks3.2.attn.proj_drop, blocks3.2.attn.qkv, blocks3.2.drop_path, blocks3.2.mlp, blocks3.2.mlp.act, blocks3.2.mlp.drop, blocks3.2.mlp.fc1, blocks3.2.mlp.fc2, blocks3.2.norm1, blocks3.2.norm2, blocks3.2.pos_embed, blocks3.3, blocks3.3.attn, blocks3.3.attn.attn_drop, blocks3.3.attn.proj, blocks3.3.attn.proj_drop, blocks3.3.attn.qkv, blocks3.3.drop_path, blocks3.3.mlp, blocks3.3.mlp.act, blocks3.3.mlp.drop, blocks3.3.mlp.fc1, blocks3.3.mlp.fc2, blocks3.3.norm1, blocks3.3.norm2, blocks3.3.pos_embed, blocks3.4.drop_path, blocks3.5.drop_path, blocks3.6.drop_path, blocks3.7.drop_path, blocks4.0.drop_path, blocks4.1.drop_path, blocks4.2.drop_path
[11/17 12:04:46][INFO] misc.py: 191: Activations: 65.24801599999999 M
[11/17 12:04:46][INFO] misc.py: 196: nvidia-smi
Thu Nov 17 12:04:46 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 522.25 Driver Version: 522.25 CUDA Version: 11.8 |
|-------------------------------+----------------------+----------------------+
| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce ... WDDM | 00000000:26:00.0 On | N/A |
| 0% 51C P2 32W / 120W | 2494MiB / 6144MiB | 9% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 1456 C+G ...\PowerToys.FancyZones.exe N/A |
| 0 N/A N/A 5844 C+G ...werToys.PowerLauncher.exe N/A |
| 0 N/A N/A 9248 C+G ...6.0.3.0\GoogleDriveFS.exe N/A |
| 0 N/A N/A 10196 C+G C:\Windows\explorer.exe N/A |
| 0 N/A N/A 11424 C+G ...ropbox\Client\Dropbox.exe N/A |
| 0 N/A N/A 11960 C+G ...8bbwe\Microsoft.Notes.exe N/A |
| 0 N/A N/A 12932 C+G ...5n1h2txyewy\SearchApp.exe N/A |
| 0 N/A N/A 13532 C+G ...bbwe\Microsoft.Photos.exe N/A |
| 0 N/A N/A 13536 C+G ...me\Application\chrome.exe N/A |
| 0 N/A N/A 15480 C+G ...2txyewy\TextInputHost.exe N/A |
| 0 N/A N/A 16648 C+G ...ram Files\LGHUB\lghub.exe N/A |
| 0 N/A N/A 17376 C+G ...perience\NVIDIA Share.exe N/A |
| 0 N/A N/A 20416 C+G ...ons\Grammarly.Desktop.exe N/A |
| 0 N/A N/A 21528 C ...envs\uniformer\python.exe N/A |
| 0 N/A N/A 22376 C+G ...werToys.ColorPickerUI.exe N/A |
| 0 N/A N/A 24568 C+G ...y\ShellExperienceHost.exe N/A |
| 0 N/A N/A 26792 C+G ...zpdnekdrzrea0\Spotify.exe N/A |
| 0 N/A N/A 28376 C+G ...icrosoft VS Code\Code.exe N/A |
+-----------------------------------------------------------------------------+
[11/17 12:04:46][INFO] checkpoint.py: 213: Loading network weights from ./exp/uniformer_s8x8_k400\checkpoints\checkpoint_epoch_00002.pyth.
[11/17 12:04:46][INFO] kinetics.py: 76: Constructing Kinetics test...
Traceback (most recent call last):
File "C:\Users\Admin\Documents\master\UniFormer\video_classification\tools\run_net.py", line 31, in <module>
main()
File "C:\Users\Admin\Documents\master\UniFormer\video_classification\tools\run_net.py", line 27, in main
launch_job(cfg=cfg, init_method=args.init_method, func=test)
File "e:\master\uniformer\video_classification\slowfast\utils\misc.py", line 296, in launch_job
torch.multiprocessing.spawn(
File "D:\CAIDATPHANMEM\miniconda3\envs\uniformer\lib\site-packages\torch\multiprocessing\spawn.py", line 230, in spawn
return start_processes(fn, args, nprocs, join, daemon, start_method='spawn')
File "D:\CAIDATPHANMEM\miniconda3\envs\uniformer\lib\site-packages\torch\multiprocessing\spawn.py", line 188, in start_processes
while not context.join():
File "D:\CAIDATPHANMEM\miniconda3\envs\uniformer\lib\site-packages\torch\multiprocessing\spawn.py", line 150, in join
raise ProcessRaisedException(msg, error_index, failed_process.pid)
torch.multiprocessing.spawn.ProcessRaisedException:
-- Process 0 terminated with the following error:
Traceback (most recent call last):
File "D:\CAIDATPHANMEM\miniconda3\envs\uniformer\lib\site-packages\torch\multiprocessing\spawn.py", line 59, in _wrap
fn(i, *args)
File "e:\master\uniformer\video_classification\slowfast\utils\multiprocessing.py", line 60, in run
ret = func(cfg)
File "C:\Users\Admin\Documents\master\UniFormer\video_classification\tools\test_net.py", line 168, in test
test_loader = loader.construct_loader(cfg, "test")
File "e:\master\uniformer\video_classification\slowfast\datasets\loader.py", line 112, in construct_loader
dataset = build_dataset(dataset_name, cfg, split)
File "e:\master\uniformer\video_classification\slowfast\datasets\build.py", line 31, in build_dataset
return DATASET_REGISTRY.get(name)(cfg, split)
File "e:\master\uniformer\video_classification\slowfast\datasets\kinetics.py", line 77, in __init__
self._construct_loader()
File "e:\master\uniformer\video_classification\slowfast\datasets\kinetics.py", line 121, in _construct_loader
self._split_idx, path_to_file
File "D:\CAIDATPHANMEM\miniconda3\envs\uniformer\lib\site-packages\torch\utils\data\dataset.py", line 83, in __getattr__
raise AttributeError
AttributeError
(uniformer)
and this is my config in run.sh
work_path=$(dirname $0)
PYTHONPATH=$PYTHONPATH:./slowfast \
python tools/run_net.py \
--cfg $work_path/config.yaml \
DATA.PATH_TO_DATA_DIR ./data_vid \
DATA.PATH_LABEL_SEPARATOR "," \
TRAIN.EVAL_PERIOD 5 \
TRAIN.CHECKPOINT_PERIOD 1 \
TRAIN.BATCH_SIZE 8 \
TRAIN.CHECKPOINT_FILE_PATH ./exp/uniformer_s8x8_k400/checkpoints/checkpoint_epoch_00002.pyth \
NUM_GPUS 1 \
UNIFORMER.DROP_DEPTH_RATE 0.1 \
SOLVER.MAX_EPOCH 2 \
SOLVER.BASE_LR 4e-4 \
SOLVER.WARMUP_EPOCHS 10.0 \
DATA.TEST_CROP_SIZE 224 \
TEST.NUM_ENSEMBLE_VIEWS 1 \
TEST.NUM_SPATIAL_CROPS 1 \
RNG_SEED 6666 \
OUTPUT_DIR $work_path
Thank you in advance!