{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/arxiv-2604-07848","title":"Information-Theoretic Requirements for Gradient-Based Task Affinity Estimation in Multi-Task Learning","arxiv_id":"2604.07848","date":"2026-04-09","proceeding":null,"authors":["Jasper Zhang","Bryan Cheng"],"abstract":"Multi-task learning shows strikingly inconsistent results -- sometimes joint training helps substantially, sometimes it actively harms performance -- yet the field lacks a principled framework for predicting these outcomes. We identify a fundamental but unstated assumption underlying gradient-based task analysis: tasks must share training instances for gradient conflicts to reveal genuine relationships. When tasks are measured on the same inputs, gradient alignment reflects shared mechanistic structure; when measured on disjoint inputs, any apparent signal conflates task relationships with distributional shift. We discover this sample overlap requirement exhibits a sharp phase transition: below 30% overlap, gradient-task correlations are statistically indistinguishable from noise; above 40%, they reliably recover known biological structure. Comprehensive validation across multiple datasets achieves strong correlations and recovers biological pathway organization. Standard benchmarks systematically violate this requirement -- MoleculeNet operates at <5% overlap, TDC at 8-14% -- far below the threshold where gradient analysis becomes meaningful. This provides the first principled explanation for seven years of inconsistent MTL results.","url_abs":"https://arxiv.org/abs/2604.07848","url_pdf":"https://arxiv.org/pdf/2604.07848","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2604.07848","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2604.07848"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/JasperZG/gradientmtl","reach":null}],"summary":{"unverified":13},"by_repo_kind":{"found_in_text":{"samples":13,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"e3895c4f86c12476","entry":"bootstrap_confidence_interval","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/statistical_tests.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/statistical_tests.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"e3895c4f86c12476"}},{"code_sha256_prefix":"9268858c0d79576b","entry":"compute_empirical_correlations","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/empirical_correlations.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/empirical_correlations.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"9268858c0d79576b"}},{"code_sha256_prefix":"010a6e50dcfe05e8","entry":"compute_empirical_matrix_tox21","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/empirical_correlations.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/empirical_correlations.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"010a6e50dcfe05e8"}},{"code_sha256_prefix":"a5a27ba424bf8c95","entry":"compute_gradient_similarity_qm9","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"experiments/cross_domain_validation.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/experiments/cross_domain_validation.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"a5a27ba424bf8c95"}},{"code_sha256_prefix":"2f789f5c396d6e36","entry":"compute_silhouette_score","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/statistical_tests.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/statistical_tests.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"2f789f5c396d6e36"}},{"code_sha256_prefix":"022c496e4b6c17ab","entry":"create_gnn_encoder","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"models/gnn_encoder.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/models/gnn_encoder.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"022c496e4b6c17ab"}},{"code_sha256_prefix":"79e573f63c0088fe","entry":"fit_sigmoid","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"experiments/cross_domain_validation.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/experiments/cross_domain_validation.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"79e573f63c0088fe"}},{"code_sha256_prefix":"6d0b32023d4337fb","entry":"get_empirical_matrix_for_tasks","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/empirical_correlations.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/empirical_correlations.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"6d0b32023d4337fb"}},{"code_sha256_prefix":"984288421ae56814","entry":"get_literature_matrix","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/literature_matrix.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/literature_matrix.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"984288421ae56814"}},{"code_sha256_prefix":"60210e7e92e2a3ea","entry":"permutation_test","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/statistical_tests.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/statistical_tests.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"60210e7e92e2a3ea"}},{"code_sha256_prefix":"dd392bf580ac3e08","entry":"sigmoid","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"experiments/cross_domain_validation.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/experiments/cross_domain_validation.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"dd392bf580ac3e08"}},{"code_sha256_prefix":"1f9e668a11cdd14b","entry":"train_epoch","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"experiments/deep_analysis_1_gradient_dynamics.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/experiments/deep_analysis_1_gradient_dynamics.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1f9e668a11cdd14b"}},{"code_sha256_prefix":"7286042fe3310563","entry":"validate_gradient_matrix","repo":"JasperZG/gradientmtl","repo_kind":"found_in_text","path":"analysis/literature_matrix.py","file_url":"https://github.com/JasperZG/gradientmtl/blob/HEAD/analysis/literature_matrix.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7286042fe3310563"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}