{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:562ZKPUS3BBOPEWV6H4JRQ6KXX","short_pith_number":"pith:562ZKPUS","schema_version":"1.0","canonical_sha256":"efb5953e92d842e792d5f1f898c3cabde050066a977a82f7538b695bcc1e6304","source":{"kind":"arxiv","id":"2607.17624","version":1},"attestation_state":"computed","paper":{"title":"Can Transformers Really Do It All? On the Compatibility of Inductive Biases Across Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Damien Teney, Hemanth Saratchandran, Liangze Jiang, Simon Lucey","submitted_at":"2026-07-20T07:26:17Z","abstract_excerpt":"Transformers are remarkably versatile and their design is largely consistent across a variety of applications. But are they optimal for any given task or dataset? The answer may be key for pushing AI beyond merely scaling current designs.\n  *Method.* We present a method to optimize a transformer architecture for a given dataset, which we use as a tool to study optimal task-specific inductive biases. This method replaces the most important non-linearities (GeLUs,;softmax) with functions learned on held-out data. We then train the resulting architectures on other datasets, as a way to evaluate t"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2607.17624","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-20T07:26:17Z","cross_cats_sorted":[],"title_canon_sha256":"cccb52e5563f712ce1dc31ceacf19504d66df7499b94c80fd46b1ec6efcaf2ac","abstract_canon_sha256":"9f7ece5038778b91fae5b8fa4dee8df54d7ebb99b52d4dd29b0d2292a28790f3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-21T01:21:55.980156Z","signature_b64":"UjIorYFCOUfVkQ68Dd0Kz1FlLCmSUqxl6IV7jNozgIxNEBBgYFUYzmNYXUalJDC2NBGguYfazGv9v5Bmc4Z7AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"efb5953e92d842e792d5f1f898c3cabde050066a977a82f7538b695bcc1e6304","last_reissued_at":"2026-07-21T01:21:55.979349Z","signature_status":"signed_v1","first_computed_at":"2026-07-21T01:21:55.979349Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can Transformers Really Do It All? On the Compatibility of Inductive Biases Across Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Damien Teney, Hemanth Saratchandran, Liangze Jiang, Simon Lucey","submitted_at":"2026-07-20T07:26:17Z","abstract_excerpt":"Transformers are remarkably versatile and their design is largely consistent across a variety of applications. But are they optimal for any given task or dataset? The answer may be key for pushing AI beyond merely scaling current designs.\n  *Method.* We present a method to optimize a transformer architecture for a given dataset, which we use as a tool to study optimal task-specific inductive biases. This method replaces the most important non-linearities (GeLUs,;softmax) with functions learned on held-out data. We then train the resulting architectures on other datasets, as a way to evaluate t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.17624","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2607.17624/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2607.17624","created_at":"2026-07-21T01:21:55.979754+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.17624v1","created_at":"2026-07-21T01:21:55.979754+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.17624","created_at":"2026-07-21T01:21:55.979754+00:00"},{"alias_kind":"pith_short_12","alias_value":"562ZKPUS3BBO","created_at":"2026-07-21T01:21:55.979754+00:00"},{"alias_kind":"pith_short_16","alias_value":"562ZKPUS3BBOPEWV","created_at":"2026-07-21T01:21:55.979754+00:00"},{"alias_kind":"pith_short_8","alias_value":"562ZKPUS","created_at":"2026-07-21T01:21:55.979754+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX","json":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX.json","graph_json":"https://pith.science/api/pith-number/562ZKPUS3BBOPEWV6H4JRQ6KXX/graph.json","events_json":"https://pith.science/api/pith-number/562ZKPUS3BBOPEWV6H4JRQ6KXX/events.json","paper":"https://pith.science/paper/562ZKPUS"},"agent_actions":{"view_html":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX","download_json":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX.json","view_paper":"https://pith.science/paper/562ZKPUS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.17624&json=true","fetch_graph":"https://pith.science/api/pith-number/562ZKPUS3BBOPEWV6H4JRQ6KXX/graph.json","fetch_events":"https://pith.science/api/pith-number/562ZKPUS3BBOPEWV6H4JRQ6KXX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX/action/storage_attestation","attest_author":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX/action/author_attestation","sign_citation":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX/action/citation_signature","submit_replication":"https://pith.science/pith/562ZKPUS3BBOPEWV6H4JRQ6KXX/action/replication_record"}},"created_at":"2026-07-21T01:21:55.979754+00:00","updated_at":"2026-07-21T01:21:55.979754+00:00"}