Open martinheidegger opened 2 years ago
During the last training process, we learned that the expensive ec2 instances we use for training are being used for pre-computation (see #5). The last execution results are stored in following tradle s3 bucket:
https://s3.console.aws.amazon.com/s3/buckets/kycdeepface-training?region=ap-east-1&prefix=models/&showversions=false
They can be used to continue training with a newly set up ec2 instance and needs to resume from by changing the start_epoch below...
https://github.com/tradle/KYCDeepFace/blob/233f33d62daa6dd4eeae9433cab58822f8208a0d/train.py#L43
... to 22 and restart the process.
22
During the last training process, we learned that the expensive ec2 instances we use for training are being used for pre-computation (see #5). The last execution results are stored in following tradle s3 bucket:
https://s3.console.aws.amazon.com/s3/buckets/kycdeepface-training?region=ap-east-1&prefix=models/&showversions=false
They can be used to continue training with a newly set up ec2 instance and needs to resume from by changing the start_epoch below...
https://github.com/tradle/KYCDeepFace/blob/233f33d62daa6dd4eeae9433cab58822f8208a0d/train.py#L43
... to
22and restart the process.