{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:MLYIFTGRHA5EJ2ZH3YKVOJFQNB","short_pith_number":"pith:MLYIFTGR","schema_version":"1.0","canonical_sha256":"62f082ccd1383a44eb27de155724b06846efd5a1f45b84fd31caab2271bc7204","source":{"kind":"arxiv","id":"2109.04947","version":3},"attestation_state":"computed","paper":{"title":"Tiered Reasoning for Intuitive Physics: Toward Verifiable Commonsense Language Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Joyce Chai, Qiaozi Gao, Shane Storks, Yichi Zhang","submitted_at":"2021-09-10T15:47:22Z","abstract_excerpt":"Large-scale, pre-trained language models (LMs) have achieved human-level performance on a breadth of language understanding tasks. However, evaluations only based on end task performance shed little light on machines' true ability in language understanding and reasoning. In this paper, we highlight the importance of evaluating the underlying reasoning process in addition to end performance. Toward this goal, we introduce Tiered Reasoning for Intuitive Physics (TRIP), a novel commonsense reasoning dataset with dense annotations that enable multi-tiered evaluation of machines' reasoning process."},"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":"2109.04947","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-09-10T15:47:22Z","cross_cats_sorted":[],"title_canon_sha256":"e818fc1c62ac947c8bc038a360e2cca7b6b4225f1e294a5eeac875db5b001c2d","abstract_canon_sha256":"bb98b94558c8ecbc30a5e2c8dc95c10c6c9a7b23f8e3b6fa2a39f106cc9fc70e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:21:56.720455Z","signature_b64":"OFn0hV77Y2bO2TKtdDmPOs2eQXXrYLgfMjwZhf4lLyyKN4tiKwXSM8+TqLNToSZwuDcgUhKoa7hgQxR5UTi4CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"62f082ccd1383a44eb27de155724b06846efd5a1f45b84fd31caab2271bc7204","last_reissued_at":"2026-07-05T04:21:56.719894Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:21:56.719894Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tiered Reasoning for Intuitive Physics: Toward Verifiable Commonsense Language Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Joyce Chai, Qiaozi Gao, Shane Storks, Yichi Zhang","submitted_at":"2021-09-10T15:47:22Z","abstract_excerpt":"Large-scale, pre-trained language models (LMs) have achieved human-level performance on a breadth of language understanding tasks. However, evaluations only based on end task performance shed little light on machines' true ability in language understanding and reasoning. In this paper, we highlight the importance of evaluating the underlying reasoning process in addition to end performance. Toward this goal, we introduce Tiered Reasoning for Intuitive Physics (TRIP), a novel commonsense reasoning dataset with dense annotations that enable multi-tiered evaluation of machines' reasoning process."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.04947","kind":"arxiv","version":3},"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/2109.04947/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":"2109.04947","created_at":"2026-07-05T04:21:56.719967+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.04947v3","created_at":"2026-07-05T04:21:56.719967+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.04947","created_at":"2026-07-05T04:21:56.719967+00:00"},{"alias_kind":"pith_short_12","alias_value":"MLYIFTGRHA5E","created_at":"2026-07-05T04:21:56.719967+00:00"},{"alias_kind":"pith_short_16","alias_value":"MLYIFTGRHA5EJ2ZH","created_at":"2026-07-05T04:21:56.719967+00:00"},{"alias_kind":"pith_short_8","alias_value":"MLYIFTGR","created_at":"2026-07-05T04:21:56.719967+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/MLYIFTGRHA5EJ2ZH3YKVOJFQNB","json":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB.json","graph_json":"https://pith.science/api/pith-number/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/graph.json","events_json":"https://pith.science/api/pith-number/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/events.json","paper":"https://pith.science/paper/MLYIFTGR"},"agent_actions":{"view_html":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB","download_json":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB.json","view_paper":"https://pith.science/paper/MLYIFTGR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.04947&json=true","fetch_graph":"https://pith.science/api/pith-number/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/graph.json","fetch_events":"https://pith.science/api/pith-number/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/action/storage_attestation","attest_author":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/action/author_attestation","sign_citation":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/action/citation_signature","submit_replication":"https://pith.science/pith/MLYIFTGRHA5EJ2ZH3YKVOJFQNB/action/replication_record"}},"created_at":"2026-07-05T04:21:56.719967+00:00","updated_at":"2026-07-05T04:21:56.719967+00:00"}