search

abhineet123 / ipsc_prediction

IPSC Prediction with Deep Learning
1 stars 0 forks source link

About training time spent #6

Closed SZ-qing closed 1 year ago

SZ-qing commented 1 year ago

Hi, I want to repeat your article, and I would like to ask you how long you trained for each part of static images and videos.

abhineet123 commented 1 year ago

Sorry, I'm not sure I entirely understand your question.

If you're asking about how many epochs each model was trained for, you can find this info in the name of the checkpoint.

For example, early and late-stage SWD models were trained for 3273 and 2058 epochs and their checkpoints are therefore named epoch_3273.pth and epoch_2058.pth respectively

SZ-qing commented 1 year ago

What I want to ask is how long does it take to train the eight models based on your computing resources and parameter settings

---- Replied Message ---- | From | Abhineet @.> | | Date | 08/11/2023 03:30 | | To | abhineet123/ipsc_prediction @.> | | Cc | SZ-qing @.>, Author @.> | | Subject | Re: [abhineet123/ipsc_prediction] About training time spent (Issue #6) |

Sorry, I'm not sure I entirely understand your question.

If you're asking about how many epochs each model was trained for, you can find this info in the name of the checkpoint.

For example, early and late-stage SWD models were trained for 3273 and 2058 epochs and their checkpoints are therefore named epoch_3273.pth and epoch_2058.pth respectively

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

abhineet123 commented 1 year ago

That is hard to say but I'd guess at an average of 1 to 2 weeks per model.

The classifiers were the fastest to train and reached convergence in just 3 - 4 days while the video segmentation models took the longest.

SZ-qing commented 1 year ago

Thanks. ---- Replied Message ---- | From | Abhineet @.> | | Date | 08/11/2023 03:48 | | To | @.> | | Cc | SZ-qing @.>, Author @.> | | Subject | Re: [abhineet123/ipsc_prediction] About training time spent (Issue #6) |

That is hard to say but I'd guess at an average of 1 to 2 weeks per model.

The classifiers were the fastest to train and reached convergence in just 3 - 4 days while the video segmentation models took the longest.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>