SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2505.23883 · NeurIPS · 2025

BIOCLIP 2: Emergent Properties from Scaling Hierarchical Contrastive Learning

Wei-Lun Chao, Yu Su, Samuel Stevens, Zheda Mai, Tanya Berger-Wolf, Daniel Rubenstein, Jianyang Gu, Wasila Dahdul, Matthew Thompson, Elizabeth Campolongo, Hilmar Lapp, Net Zhang, and 4 more

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 4 functions out of this paper's own repositories and ran 4 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
imageomics/treeoflife-toolbox canonical 3 of 3
Imageomics/bioclip-2 canonical 1 of 1
FunctionStatusWhere it lives
get_id Ran imageomics/treeoflife-toolbox/src/TreeOfLife_toolbox/main/utils.py
code served (permissive licence) · get_code("8efb8d787770bcad")
init_logger Ran imageomics/treeoflife-toolbox/src/TreeOfLife_toolbox/main/utils.py
code served (permissive licence) · get_code("2edc98b39c0ff429")
is_subset Ran imageomics/treeoflife-toolbox/src/TreeOfLife_toolbox/main/config.py
code served (permissive licence) · get_code("15fc5ea70b4ee157")
text_global_pool Ran Imageomics/bioclip-2/src/open_clip/transformer.py
pointer only (licence: NOASSERTION) · get_code("c585d91a8dca5374")

Repositories linked to this paper

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

Abstract

Female Juvenile Juvenile Adult Adult Male Male Female Species span Figure 1: While BIOCLIP 2 is trained to distinguish species, it demonstrates emergent properties beyond the initial training objective. Left: At the inter-species level, the embedding distribution of different species aligns with ecological relationships; the embeddings of Darwin's finches arrange themselves by beak size from left to right. Right: Instead of collapsing, the intra-species variations are preserved in subspaces orthogonal to the inter-species variation (the black lines point from the mean embedding of one variant to that of the other variant). Orthogonality increases with scale (see Figure 3c).

For agents

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

get_harvested_code_for_paper("2505.23883")
get_code_for_paper("2505.23883")
have("2505.23883")

Connect an agent — have() is free.