[KongchangAI]
ConceptValidation Loss

Validation Loss

A metric in machine learning that measures the discrepancy between a model's predicted probability distribution and the true labels. It is commonly used to monitor the training process and to select checkpoints.

Timeline (last 90 days)

Nov 11

Validation loss (such as cross-entropy) measures the difference between the model's predicted probability distribution and the true labels, while validation accuracy (or F1) measures the correctness of hard classification decisions; the two are not always monotonically in sync.

Unverified50%
Nov 11

The checkpoint selected by the lowest-loss criterion and the checkpoint selected by the highest-accuracy criterion may be models from different stages of training, so their test performance may naturally differ.

Unverified50%
Oct 21

In the DPO setting, there is only a weak correlation between the hyperparameters that minimize validation loss and the optimal hyperparameters on downstream benchmarks.

Unverified50%
Oct 21

Using validation loss for early stopping or model selection may point in the wrong direction in the DPO setting.

Unverified50%
Oct 21

The validation loss on preference data measures how well the model fits the preferred responses, whereas downstream benchmarks examine generation quality; the two are not equivalent, and the model may reduce validation loss while degrading on downstream tasks.

Unverified50%
Oct 17

PPO and DPO have different objective functions, so validation loss cannot be compared across them; perplexity loses meaning after PPO.

Unverified50%

All Facts (6)

Source Articles