LaMP-2
The movie tag personalization task within LaMP, a benchmark for language model personalization. It requires the model to predict personalized tags based on a user's historical movie reviews and tags, and is commonly used to evaluate the effectiveness of on-device personalized fine-tuning.
Timeline (last 90 days)
The workloads cover a nighttime personalization task based on phone-provided labels and the public LaMP-2 benchmark.
LaMP-2 is the movie-tag personalization task in the LaMP benchmark: given a user's historical movie reviews and their tags, the model is required to predict personalized tags for the user's new movie review.
On 200 users of the LaMP-2 dataset (a 15-category news classification task), PersonaLink achieves an accuracy of 0.745–0.755.
On the LaMP-2 dataset, the BM25-based retrieval method achieves an accuracy of 0.760–0.765, which is statistically indistinguishable from PersonaLink.
All Facts (4)
The workloads cover a nighttime personalization task based on phone-provided labels and the public LaMP-2 benchmark.
50%UnverifiedLaMP-2 is the movie-tag personalization task in the LaMP benchmark: given a user's historical movie reviews and their tags, the model is required to predict personalized tags for the user's new movie review.
50%UnverifiedOn 200 users of the LaMP-2 dataset (a 15-category news classification task), PersonaLink achieves an accuracy of 0.745–0.755.
50%UnverifiedOn the LaMP-2 dataset, the BM25-based retrieval method achieves an accuracy of 0.760–0.765, which is statistically indistinguishable from PersonaLink.
50%