When I want to train the model by run the below command:
python train.py --data_dir /home/whw/Tracking/Pytorch-SiamFC/ILSVRC --exp_name fc -f safe
The output is listed as follow:
[IMAGE-UTILS] package jpeg4py not available. Continuing...
----Starting train script in mode: train----
Loading datasets...
Validation dataset...
Metadata file found. Loading its content.
Training dataset...
Metadata file found. Loading its content.
The user set train_epoch_size (10000) is bigger than the size of the train set (6280).
Setting train_epoch_size to the train set size.
Done
Setup time: 0h0m1.56s
Using Exponential Learning Rate Decay of 1
Epoch sizes: 6280 in train and 10000 in eval
Starting training for 1 epoch(s)
Pretraining evaluation...
Validation on val set
0%| | 0/10000 [00:00<?, ?it/s]Saving embeddings for summary 0
Saving embeddings for summary 1
Saving embeddings for summary 2
Saving embeddings for summary 3
Saving embeddings for summary 4
0%| | 0/10000 [00:03<?, ?it/s]
Fatal error in main loop
Traceback (most recent call last):
File "train.py", line 193, in main
args, summ_maker=summ_maker)
File "train.py", line 271, in train_and_evaluate
summ_maker=summ_maker)
File "/home/whw/anaconda3/envs/siamfc/lib/python3.6/site-packages/torch/autograd/grad_mode.py", line 46, in decorate_no_grad
return func(*args, **kwargs)
File "train.py", line 496, in evaluate
for metric_name, metric_dict in metrics.items()}
File "train.py", line 496, in
for metric_name, metric_dict in metrics.items()}
File "/home/whw/Tracking/Pytorch-SiamFC/training/metrics.py", line 60, in AUC
if np.isnan(label[i]) or np.isnan(output[i]):
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
=== Execution Terminated with error ===
Who can help me?
hwmaybe
commented
4 years ago
When I want to train the model by run the below command: python train.py --data_dir /home/whw/Tracking/Pytorch-SiamFC/ILSVRC --exp_name fc -f safe The output is listed as follow: [IMAGE-UTILS] package jpeg4py not available. Continuing... ----Starting train script in mode: train---- Loading datasets... Validation dataset... Metadata file found. Loading its content. Training dataset... Metadata file found. Loading its content. The user set train_epoch_size (10000) is bigger than the size of the train set (6280). Setting train_epoch_size to the train set size. Done Setup time: 0h0m1.56s Using Exponential Learning Rate Decay of 1 Epoch sizes: 6280 in train and 10000 in eval Starting training for 1 epoch(s) Pretraining evaluation... Validation on val set 0%| | 0/10000 [00:00<?, ?it/s]Saving embeddings for summary 0 Saving embeddings for summary 1 Saving embeddings for summary 2 Saving embeddings for summary 3 Saving embeddings for summary 4 0%| | 0/10000 [00:03<?, ?it/s] Fatal error in main loop Traceback (most recent call last): File "train.py", line 193, in main args, summ_maker=summ_maker) File "train.py", line 271, in train_and_evaluate summ_maker=summ_maker) File "/home/whw/anaconda3/envs/siamfc/lib/python3.6/site-packages/torch/autograd/grad_mode.py", line 46, in decorate_no_grad return func(*args, **kwargs) File "train.py", line 496, in evaluate for metric_name, metric_dict in metrics.items()} File "train.py", line 496, in
for metric_name, metric_dict in metrics.items()}
File "/home/whw/Tracking/Pytorch-SiamFC/training/metrics.py", line 60, in AUC
if np.isnan(label[i]) or np.isnan(output[i]):
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
=== Execution Terminated with error ===
Who can help me?