SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.02465 · 2024

An Empirical Study into Clustering of Unseen Datasets with Self-Supervised Encoders

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 18 functions out of this paper's own repositories and ran 15 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
scottclowe/zs-ssl-clustering canonical 15 of 18
FunctionStatusWhere it lives
build_affinity_matrix Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/louvain.py
code served (permissive licence) · get_code("f08f66e56561b459")
determine_epoch_seed Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/utils.py
code served (permissive licence) · get_code("da25bbcd7de8221b")
generate_id Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/utils.py
code served (permissive licence) · get_code("1bf2e9396bc00670")
get_continuous_class_map Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/datasets/nabirds.py
code served (permissive licence) · get_code("d7aa9c6a98fa4fc5")
get_embeddings_path Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/io.py
code served (permissive licence) · get_code("f3d48f6c34f39530")
get_pred_path Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/io.py
code served (permissive licence) · get_code("f96c64779258568e")
get_randsizecrop_transform Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/data_transformations.py
code served (permissive licence) · get_code("4ebdf4346a9ed2e7")
get_transform Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/data_transformations.py
code served (permissive licence) · get_code("c0cd4e224662e2f7")
image_dataset_sizes Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/datasets/api.py
code served (permissive licence) · get_code("2252d616ba4317bf")
load_class_names Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/datasets/nabirds.py
code served (permissive licence) · get_code("d53209ecea508922")
load_hierarchy Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/datasets/nabirds.py
code served (permissive licence) · get_code("02112df93daa57bf")
probe_embedding_shape Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/probe.py
code served (permissive licence) · get_code("c9646fb0b16f0e8b")
sanitize_filename Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/io.py
code served (permissive licence) · get_code("3a3579dedf3fab37")
scale_lr Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/probe.py
code served (permissive licence) · get_code("3b2cc66fa66500b3")
setup_linear_classifiers Ran scottclowe/zs-ssl-clustering/zs_ssl_clustering/probe.py
code served (permissive licence) · get_code("506400c268aaa688")
fetch_dataset Not yet run scottclowe/zs-ssl-clustering/zs_ssl_clustering/datasets/api.py
code served (permissive licence) · get_code("d0fc1c7854a8ad46")
fetch_image_dataset Not yet run scottclowe/zs-ssl-clustering/zs_ssl_clustering/datasets/api.py
code served (permissive licence) · get_code("94ded4fcf407632d")
init_or_resume_wandb_run Not yet run scottclowe/zs-ssl-clustering/zs_ssl_clustering/utils.py
code served (permissive licence) · get_code("c5b56f5bbb6ef6cd")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Can pretrained models generalize to new datasets without any retraining? We deploy pretrained image models on datasets they were not trained for, and investigate whether their embeddings form meaningful clusters. Our suite of benchmarking experiments use encoders pretrained solely on ImageNet-1k with either supervised or self-supervised training techniques, deployed on image datasets that were not seen during training, and clustered with conventional clustering algorithms. This evaluation provides new insights into the embeddings of self-supervised models, which prioritize different features to supervised models. Supervised encoders typically offer more utility than SSL encoders within the training domain, and vice-versa far outside of it, however, fine-tuned encoders demonstrate the opposite trend. Clustering provides a way to evaluate the utility of self-supervised learned representations orthogonal to existing methods such as kNN. Additionally, we find the silhouette score when measured in a UMAP-reduced space is highly correlated with clustering performance, and can therefore be used as a proxy for clustering performance on data with no ground truth labels. Our code implementation is available at \url{https://github.com/scottclowe/zs-ssl-clustering/}.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2406.02465")
get_code_for_paper("2406.02465")
have("2406.02465")

Connect an agent — have() is free.