{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2020:QCAXLNMOABZIDNI5YRPFBIP4OF","short_pith_number":"pith:QCAXLNMO","canonical_record":{"source":{"id":"2003.07892","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-03-17T18:58:44Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"8bc2b539cb085dbb1d4cde776ce2f69155ef7320bd24584dc9cbc2c48be7d70b","abstract_canon_sha256":"cb758587801e7ae8889be271905d13a8c28e6ffff02a8109c1c5dacf9c062ba8"},"schema_version":"1.0"},"canonical_sha256":"808175b58e007281b51dc45e50a1fc716704bff81c89e98fa669239819cc5fdf","source":{"kind":"arxiv","id":"2003.07892","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2003.07892","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"arxiv_version","alias_value":"2003.07892v3","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2003.07892","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"pith_short_12","alias_value":"QCAXLNMOABZI","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"pith_short_16","alias_value":"QCAXLNMOABZIDNI5","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"pith_short_8","alias_value":"QCAXLNMO","created_at":"2026-07-05T01:43:14Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2020:QCAXLNMOABZIDNI5YRPFBIP4OF","target":"record","payload":{"canonical_record":{"source":{"id":"2003.07892","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-03-17T18:58:44Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"8bc2b539cb085dbb1d4cde776ce2f69155ef7320bd24584dc9cbc2c48be7d70b","abstract_canon_sha256":"cb758587801e7ae8889be271905d13a8c28e6ffff02a8109c1c5dacf9c062ba8"},"schema_version":"1.0"},"canonical_sha256":"808175b58e007281b51dc45e50a1fc716704bff81c89e98fa669239819cc5fdf","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:43:14.432308Z","signature_b64":"RrLKlIhgU9XArwItCNdQkjJdrG5OMpIEV47zHPbsQ6s5D6IuTQPYK6Kq9CTSVNEjEbG9c1C/Npl5qqQDF2F1BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"808175b58e007281b51dc45e50a1fc716704bff81c89e98fa669239819cc5fdf","last_reissued_at":"2026-07-05T01:43:14.431847Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:43:14.431847Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2003.07892","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T01:43:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/8gNnom/dE7vTcQ8l48wK3DERuYCLMK+O26gu+Lgr9CHRT5ATtzhokSnkQOx9xDJlPQCxKZkzFDp50oyYop4Ag==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T11:45:14.990441Z"},"content_sha256":"c925460d86e34e2f1698d34c9d1d50ec47e17095e68d715538fff233cd35960c","schema_version":"1.0","event_id":"sha256:c925460d86e34e2f1698d34c9d1d50ec47e17095e68d715538fff233cd35960c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2020:QCAXLNMOABZIDNI5YRPFBIP4OF","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Calibration of Pre-trained Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Greg Durrett, Shrey Desai","submitted_at":"2020-03-17T18:58:44Z","abstract_excerpt":"Pre-trained Transformers are now ubiquitous in natural language processing, but despite their high end-task performance, little is known empirically about whether they are calibrated. Specifically, do these models' posterior probabilities provide an accurate empirical measure of how likely the model is to be correct on a given example? We focus on BERT and RoBERTa in this work, and analyze their calibration across three tasks: natural language inference, paraphrase detection, and commonsense reasoning. For each task, we consider in-domain as well as challenging out-of-domain settings, where mo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2003.07892","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/2003.07892/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T01:43:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ploc0Ouc8jYjrqaoWO/J2F39+PzodOe91FfPzRh/rnv9ziC9YIJFaplovGJ664F1wQBC5H/nVKH6X2QA3us3Cw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T11:45:14.990952Z"},"content_sha256":"bcc2bfd80c2dea04ff88ab99e93e25759495918d9ee92857c220d7526145c459","schema_version":"1.0","event_id":"sha256:bcc2bfd80c2dea04ff88ab99e93e25759495918d9ee92857c220d7526145c459"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/QCAXLNMOABZIDNI5YRPFBIP4OF/bundle.json","state_url":"https://pith.science/pith/QCAXLNMOABZIDNI5YRPFBIP4OF/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/QCAXLNMOABZIDNI5YRPFBIP4OF/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-06T11:45:14Z","links":{"resolver":"https://pith.science/pith/QCAXLNMOABZIDNI5YRPFBIP4OF","bundle":"https://pith.science/pith/QCAXLNMOABZIDNI5YRPFBIP4OF/bundle.json","state":"https://pith.science/pith/QCAXLNMOABZIDNI5YRPFBIP4OF/state.json","well_known_bundle":"https://pith.science/.well-known/pith/QCAXLNMOABZIDNI5YRPFBIP4OF/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:QCAXLNMOABZIDNI5YRPFBIP4OF","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"cb758587801e7ae8889be271905d13a8c28e6ffff02a8109c1c5dacf9c062ba8","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-03-17T18:58:44Z","title_canon_sha256":"8bc2b539cb085dbb1d4cde776ce2f69155ef7320bd24584dc9cbc2c48be7d70b"},"schema_version":"1.0","source":{"id":"2003.07892","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2003.07892","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"arxiv_version","alias_value":"2003.07892v3","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2003.07892","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"pith_short_12","alias_value":"QCAXLNMOABZI","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"pith_short_16","alias_value":"QCAXLNMOABZIDNI5","created_at":"2026-07-05T01:43:14Z"},{"alias_kind":"pith_short_8","alias_value":"QCAXLNMO","created_at":"2026-07-05T01:43:14Z"}],"graph_snapshots":[{"event_id":"sha256:bcc2bfd80c2dea04ff88ab99e93e25759495918d9ee92857c220d7526145c459","target":"graph","created_at":"2026-07-05T01:43:14Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2003.07892/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Pre-trained Transformers are now ubiquitous in natural language processing, but despite their high end-task performance, little is known empirically about whether they are calibrated. Specifically, do these models' posterior probabilities provide an accurate empirical measure of how likely the model is to be correct on a given example? We focus on BERT and RoBERTa in this work, and analyze their calibration across three tasks: natural language inference, paraphrase detection, and commonsense reasoning. For each task, we consider in-domain as well as challenging out-of-domain settings, where mo","authors_text":"Greg Durrett, Shrey Desai","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-03-17T18:58:44Z","title":"Calibration of Pre-trained Transformers"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2003.07892","kind":"arxiv","version":3},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c925460d86e34e2f1698d34c9d1d50ec47e17095e68d715538fff233cd35960c","target":"record","created_at":"2026-07-05T01:43:14Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"cb758587801e7ae8889be271905d13a8c28e6ffff02a8109c1c5dacf9c062ba8","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-03-17T18:58:44Z","title_canon_sha256":"8bc2b539cb085dbb1d4cde776ce2f69155ef7320bd24584dc9cbc2c48be7d70b"},"schema_version":"1.0","source":{"id":"2003.07892","kind":"arxiv","version":3}},"canonical_sha256":"808175b58e007281b51dc45e50a1fc716704bff81c89e98fa669239819cc5fdf","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"808175b58e007281b51dc45e50a1fc716704bff81c89e98fa669239819cc5fdf","first_computed_at":"2026-07-05T01:43:14.431847Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T01:43:14.431847Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"RrLKlIhgU9XArwItCNdQkjJdrG5OMpIEV47zHPbsQ6s5D6IuTQPYK6Kq9CTSVNEjEbG9c1C/Npl5qqQDF2F1BA==","signature_status":"signed_v1","signed_at":"2026-07-05T01:43:14.432308Z","signed_message":"canonical_sha256_bytes"},"source_id":"2003.07892","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c925460d86e34e2f1698d34c9d1d50ec47e17095e68d715538fff233cd35960c","sha256:bcc2bfd80c2dea04ff88ab99e93e25759495918d9ee92857c220d7526145c459"],"state_sha256":"dc3adc70ef40967d8841e3679743468db30fd455b10986cac94a244a735a32bc"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"MUiZwUR8eXys4JyLcx99vxl3Gf9D1LPSY6OdCA/geOkJ+eHnhSnaKUxd8mgNzaa3QuFp1/KRLTfT2z3RFeLnCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T11:45:14.994665Z","bundle_sha256":"1c1b08a672649e74c083644ebf0e2eb6721c82f6f1a30abe13cc1b8c3ec7d82b"}}