aidotse/LeakPro

Loss in training loops

開放

#239 建立於 2025年2月24日

 (1 則留言) (0 個反應) (0 位負責人)Python (26 個分叉)auto 404
good first issue

倉庫指標

星標
 (23 顆星)
PR 合併指標
 (PR 指標待抓取)

描述

Issue

Problem Description

The last batch of each epoch is usually smaller than the others. One approach is to sum the losses and divide by the total number of samples.

Currently, loss calculations in handlers sum the average loss of each batch and divide by the length of the dataloader.

Expected Behavior

What Needs to be Done

How Can It Be Tested or Reproduced

貢獻者指南