Train.py use early stoping method

andy8902 · September 20, 2020, 10:25am

I use this web(pytorch-classification/train.py at 5107de352cade326e7aacd44bf8d625b7055fd4e · dusty-nv/pytorch-classification · GitHub)to use train.py. I want to add the early stopping method to train.py. How to do it?

AastaLLL · September 21, 2020, 2:47am

Hi,

You can terminate the process once it reaches the expected accuracy directly.
Then use the --resume flag if you want to resume the training job again.

github.com

dusty-nv/pytorch-classification/blob/5107de352cade326e7aacd44bf8d625b7055fd4e/train.py#L72


      
                                   'using Data Parallel or Distributed Data Parallel')
          parser.add_argument('--lr', '--learning-rate', default=0.1, type=float,
                              metavar='LR', help='initial learning rate', dest='lr')
          parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
                              help='momentum')
          parser.add_argument('--wd', '--weight-decay', default=1e-4, type=float,
                              metavar='W', help='weight decay (default: 1e-4)',
                              dest='weight_decay')
          parser.add_argument('-p', '--print-freq', default=10, type=int,
                              metavar='N', help='print frequency (default: 10)')
          parser.add_argument('--resume', default='', type=str, metavar='PATH',
                              help='path to latest checkpoint (default: none)')
          parser.add_argument('-e', '--evaluate', dest='evaluate', action='store_true',
                              help='evaluate model on validation set')
          parser.add_argument('--pretrained', dest='pretrained', action='store_true', default=True,
                              help='use pre-trained model')
          parser.add_argument('--world-size', default=-1, type=int,
                              help='number of nodes for distributed training')
          parser.add_argument('--rank', default=-1, type=int,
                              help='node rank for distributed training')
          parser.add_argument('--dist-url', default='tcp://224.66.41.62:23456', type=str,

Thanks.

andy8902 · September 21, 2020, 4:41am

I am very sorry to say. Would it be convenient for you to show or write the early_stopping code with train.py to me?

andy8902 · September 21, 2020, 7:20am

Or how to use dropout method on train.py?

AastaLLL · September 22, 2020, 5:41am

Hi,

Dropout should be added into the model directly.
Here are some related discussion for your reference:

Thanks.

andy8902 · September 23, 2020, 10:07am

I use this URL (jetson-inference/pytorch-collect.md at master · dusty-nv/jetson-inference · GitHub) to go to the training step. How can I change the weight decay of train.py (pytorch-classification/train.py at 5107de352cade326e7aacd44bf8d625b7055fd4e · dusty-nv/pytorch-classification · GitHub) without affecting the bias value?

AastaLLL · October 14, 2020, 4:03am

Hi,

The training weight is updated via back propagation with the loss value automatically.
If you want to freeze the bias value, please try to implement a custom updater or set the trainable tag into OFF.

Thanks.

Topic		Replies	Views
Re-training ResNet-18 Model Jetson TX2 jetson-inference	6	626	October 18, 2021
Retraining data for Object Detection Jetson Nano jetson-inference	2	329	October 15, 2021
Jetson nano - train model for my own object detection Jetson Nano ai-training	11	4455	October 15, 2021
Train_ssd.py dosen't work with pascal voc dataset Jetson Nano ai-training	5	1118	February 9, 2022
Googlenet network's place Jetson Nano jetson-inference	25	1253	October 18, 2021
Googlenet on jetson nano Jetson Nano	4	467	October 14, 2021
Using re-trained model inside Python script Jetson Nano ai-training	14	1981	October 15, 2021
Retrain ssd mobilnet Jetson Nano jetson-inference , ssd , ai-training	2	1109	October 15, 2021
How to change the model in tutroial - "Re-training on the Cat/Dog Dataset" by using "--arch"? Jetson Nano ai-training	3	691	November 30, 2022
What is the proper way to do classification pretrained? Jetson Nano ai-training	5	646	July 29, 2022

Train.py use early stoping method

Related topics