{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:CZH7SXLTQN6O5H3SDQE3SQASQ3","short_pith_number":"pith:CZH7SXLT","schema_version":"1.0","canonical_sha256":"164ff95d73837cee9f721c09b9401286ff98e39b1885d5a2bfc0a572b7dc70fc","source":{"kind":"arxiv","id":"2306.00258","version":1},"attestation_state":"computed","paper":{"title":"Towards Foundation Models for Scientific Machine Learning: Characterizing Scaling and Transfer Behavior","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NA","math.NA"],"primary_cat":"cs.LG","authors_text":"Amir Gholami, Dmitriy Morozov, Kurt Keutzer, Michael Mahoney, Peter Harrington, Shashank Subramanian, Wahid Bhimji","submitted_at":"2023-06-01T00:32:59Z","abstract_excerpt":"Pre-trained machine learning (ML) models have shown great performance for a wide range of applications, in particular in natural language processing (NLP) and computer vision (CV). Here, we study how pre-training could be used for scientific machine learning (SciML) applications, specifically in the context of transfer learning. We study the transfer behavior of these models as (i) the pre-trained model size is scaled, (ii) the downstream training dataset size is scaled, (iii) the physics parameters are systematically pushed out of distribution, and (iv) how a single model pre-trained on a mix"},"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":"2306.00258","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-06-01T00:32:59Z","cross_cats_sorted":["cs.NA","math.NA"],"title_canon_sha256":"e3f4444b52e573f983f54a17dd601f4dd3b52aec0c5d2e5f9a39db3283a1f249","abstract_canon_sha256":"9fb3a984276c599ec0f87f6d4a7616adf76d11638de1866edacd8d6c5f5cc7a2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:30:03.047551Z","signature_b64":"gXW322gkt2grwZIMeH0ETAuDnytiX5228Nq3MrkB7vAH3R3YysbsLT6hSf/fmUPTvEpNkCyyr2ERFd/WoCTjDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"164ff95d73837cee9f721c09b9401286ff98e39b1885d5a2bfc0a572b7dc70fc","last_reissued_at":"2026-07-05T07:30:03.047089Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:30:03.047089Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Foundation Models for Scientific Machine Learning: Characterizing Scaling and Transfer Behavior","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NA","math.NA"],"primary_cat":"cs.LG","authors_text":"Amir Gholami, Dmitriy Morozov, Kurt Keutzer, Michael Mahoney, Peter Harrington, Shashank Subramanian, Wahid Bhimji","submitted_at":"2023-06-01T00:32:59Z","abstract_excerpt":"Pre-trained machine learning (ML) models have shown great performance for a wide range of applications, in particular in natural language processing (NLP) and computer vision (CV). Here, we study how pre-training could be used for scientific machine learning (SciML) applications, specifically in the context of transfer learning. We study the transfer behavior of these models as (i) the pre-trained model size is scaled, (ii) the downstream training dataset size is scaled, (iii) the physics parameters are systematically pushed out of distribution, and (iv) how a single model pre-trained on a mix"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.00258","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/2306.00258/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":"2306.00258","created_at":"2026-07-05T07:30:03.047148+00:00"},{"alias_kind":"arxiv_version","alias_value":"2306.00258v1","created_at":"2026-07-05T07:30:03.047148+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.00258","created_at":"2026-07-05T07:30:03.047148+00:00"},{"alias_kind":"pith_short_12","alias_value":"CZH7SXLTQN6O","created_at":"2026-07-05T07:30:03.047148+00:00"},{"alias_kind":"pith_short_16","alias_value":"CZH7SXLTQN6O5H3S","created_at":"2026-07-05T07:30:03.047148+00:00"},{"alias_kind":"pith_short_8","alias_value":"CZH7SXLT","created_at":"2026-07-05T07:30:03.047148+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.01349","citing_title":"PI-JEPA: Label-Free Surrogate Pretraining for Coupled Multiphysics Simulation via Operator-Split Latent Prediction","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11033","citing_title":"TokaMind for Power Grid: Cross-Domain Transfer from Fusion Plasma","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21101","citing_title":"A Hybridizable Neural Time Integrator for Stable Autoregressive Forecasting","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3","json":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3.json","graph_json":"https://pith.science/api/pith-number/CZH7SXLTQN6O5H3SDQE3SQASQ3/graph.json","events_json":"https://pith.science/api/pith-number/CZH7SXLTQN6O5H3SDQE3SQASQ3/events.json","paper":"https://pith.science/paper/CZH7SXLT"},"agent_actions":{"view_html":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3","download_json":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3.json","view_paper":"https://pith.science/paper/CZH7SXLT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2306.00258&json=true","fetch_graph":"https://pith.science/api/pith-number/CZH7SXLTQN6O5H3SDQE3SQASQ3/graph.json","fetch_events":"https://pith.science/api/pith-number/CZH7SXLTQN6O5H3SDQE3SQASQ3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3/action/storage_attestation","attest_author":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3/action/author_attestation","sign_citation":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3/action/citation_signature","submit_replication":"https://pith.science/pith/CZH7SXLTQN6O5H3SDQE3SQASQ3/action/replication_record"}},"created_at":"2026-07-05T07:30:03.047148+00:00","updated_at":"2026-07-05T07:30:03.047148+00:00"}