Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time

The conventional recipe for maximizing model accuracy is to (1) train multiple models with various hyperparameters and (2) pick the individual model which performs best on a held-out validation set, discarding the remainder. In this paper, we revisit the second step of this procedure in the context of fine-tuning large pre-trained models, where fine-tuned models often appear to lie in a single low error basin. We show that averaging the weights of multiple models fine-tuned with different hyperparameter configurations often improves accuracy and robustness. Unlike a conventional ensemble, we may average many models without incurring any additional inference or memory costs—we call the results “model soups.” When fine-tuning large pre-trained models such as CLIP, ALIGN, and a ViT-G pretrained on JFT, our soup recipe provides signifi-cant improvements over the best model in a hyperparameter sweep on ImageNet. The resulting ViT-G model, which attains 90.94% top-1 accuracy on ImageNet, achieved a new state of the art. Furthermore, we show that the model soup approach extends to multiple image classification and natural language processing tasks, improves out-of-distribution performance, and improves zero-shot performance on new downstream tasks. Finally, we analytically relate the performance similarity of weight-averaging and logit-ensembling to flatness of the loss and confidence of the predictions, and validate this relation empirically. Code is available at https://github. com/mlfoundations/model-soups . a CLIP and ImageNet-22k pre-trained ViT-B/32 model, we use the best hyperparameters found by the hyperparameter sweep to fine-tune multiple times, varying only the random seed. For an experimental budget of 1 ≤ k ≤ 5 models, we show: (i) the individual model with random seed k , (ii) the model soup composed of models with random seeds 1 through k , and (iii) the ensemble composed of models with random seeds 1 through k .