Skip to content

LibFewShot学习使用过程中的一些困惑。 #55

Answered by wZuck
MasterWinston asked this question in Q&A
Discussion options

You must be logged in to vote
  1. val集仍然是用来挑选最优模型,参考

    LibFewShot/core/trainer.py

    Lines 102 to 110 in f002bf3

    if self.rank == 0:
    if ((epoch_idx + 1) % self.val_per_epoch) == 0:
    if val_acc > self.best_val_acc:
    self.best_val_acc = val_acc
    self.best_test_acc = test_acc
    self._save_model(epoch_idx, SaveType.BEST)
    if epoch_idx != 0 and epoch_idx % self.config["save_interval"] == 0:
    self._save_model(epoch_idx, SaveType.NORMAL)
    ,训练阶段的text on test set只是用来更直观地感受模型训练过程,给分析调优提供信息。在现在的数据集分割中,train/val/test都不相交,你说的76:24的数据集分割策略有具体的论文可以参考吗,可以拿来再讨论。

  2. 理论上是可以的,你其实想做的是在新数据集上迁移模型并微调,一般我们会在配置文件中填写预训练好的backbone权重来微调,而不是以resume的方式来继续训练,因为这可能带来一些例如学习率、学习率调度器上的问题。当你的微调任务和预训练任务的way不同时,如果你使用的…

Replies: 7 comments 9 replies

Comment options

You must be logged in to vote
5 replies
@MasterWinston
Comment options

@wZuck
Comment options

@MasterWinston
Comment options

@MasterWinston
Comment options

@wZuck
Comment options

Answer selected by MasterWinston
Comment options

You must be logged in to vote
0 replies
Comment options

You must be logged in to vote
2 replies
@wZuck
Comment options

@yangcedrus
Comment options

Comment options

You must be logged in to vote
1 reply
@wZuck
Comment options

Comment options

You must be logged in to vote
0 replies
Comment options

You must be logged in to vote
0 replies
Comment options

You must be logged in to vote
1 reply
@wZuck
Comment options

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
4 participants