{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2021:IE5AYRLKXPWXQAKDIA7VJZFBG6","short_pith_number":"pith:IE5AYRLK","canonical_record":{"source":{"id":"2102.01386","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-02T08:35:02Z","cross_cats_sorted":[],"title_canon_sha256":"3c508e316ca55bbcfca33aa0a183f7e41673c16ccb37f5f748f2219bde0e26e0","abstract_canon_sha256":"e564a364b7e0fe99ed564846a62cc918cf01bf3be560237a4a1bbd759e1d6889"},"schema_version":"1.0"},"canonical_sha256":"413a0c456abbed780143403f54e4a1378e2522d624a09193f8ace323fa792159","source":{"kind":"arxiv","id":"2102.01386","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2102.01386","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"arxiv_version","alias_value":"2102.01386v2","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.01386","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"pith_short_12","alias_value":"IE5AYRLKXPWX","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"pith_short_16","alias_value":"IE5AYRLKXPWXQAKD","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"pith_short_8","alias_value":"IE5AYRLK","created_at":"2026-07-05T02:28:55Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2021:IE5AYRLKXPWXQAKDIA7VJZFBG6","target":"record","payload":{"canonical_record":{"source":{"id":"2102.01386","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-02T08:35:02Z","cross_cats_sorted":[],"title_canon_sha256":"3c508e316ca55bbcfca33aa0a183f7e41673c16ccb37f5f748f2219bde0e26e0","abstract_canon_sha256":"e564a364b7e0fe99ed564846a62cc918cf01bf3be560237a4a1bbd759e1d6889"},"schema_version":"1.0"},"canonical_sha256":"413a0c456abbed780143403f54e4a1378e2522d624a09193f8ace323fa792159","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:28:55.467089Z","signature_b64":"idqrmvPREKIpXgJH/77KPAvs7PmTlw6mzT6CkGxEpCkr2k1Ka8WBQbj0AocZZPuXABcmxT/zMv/BVaGlKrw8Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"413a0c456abbed780143403f54e4a1378e2522d624a09193f8ace323fa792159","last_reissued_at":"2026-07-05T02:28:55.466579Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:28:55.466579Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2102.01386","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T02:28:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ix9s/5RT/Q/95Z4PmLjbyxskunw+MJBahIWaQinpLqGBT9z/twD3XMkRLND7/XXp8HYEyAPYzW+2XLTJHY/eDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T20:24:29.266987Z"},"content_sha256":"d3363ecf58985fb10e662be084430ddd701a758fb878ab2e3660d422f7afc60f","schema_version":"1.0","event_id":"sha256:d3363ecf58985fb10e662be084430ddd701a758fb878ab2e3660d422f7afc60f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2021:IE5AYRLKXPWXQAKDIA7VJZFBG6","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"AutoFreeze: Automatically Freezing Model Blocks to Accelerate Fine-tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Saurabh Agarwal, Shivaram Venkataraman, Yuhan Liu","submitted_at":"2021-02-02T08:35:02Z","abstract_excerpt":"With the rapid adoption of machine learning (ML), a number of domains now use the approach of fine tuning models which were pre-trained on a large corpus of data. However, our experiments show that even fine-tuning on models like BERT can take many hours even when using modern accelerators like GPUs. While prior work proposes limiting the number of layers that are fine-tuned, e.g., freezing all layers but the last layer, we find that such static approaches lead to reduced accuracy. We propose, AutoFreeze, a system that uses an adaptive approach to choose which layers are trained and show how t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.01386","kind":"arxiv","version":2},"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/2102.01386/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":"2026-07-05T02:28:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"z50yIJZ0PtNfE5rcekM73gH9xFK6TVJwBecRh3OccU73mmH82VI+BCo3070ZY5PJDUsv/rUX3tzUkT/JENmdDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T20:24:29.268102Z"},"content_sha256":"8f3126bb0aeb021679d5ddbc0af0532f9209d2faa44ee8643673cd7f93555c14","schema_version":"1.0","event_id":"sha256:8f3126bb0aeb021679d5ddbc0af0532f9209d2faa44ee8643673cd7f93555c14"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IE5AYRLKXPWXQAKDIA7VJZFBG6/bundle.json","state_url":"https://pith.science/pith/IE5AYRLKXPWXQAKDIA7VJZFBG6/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IE5AYRLKXPWXQAKDIA7VJZFBG6/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":"2026-08-09T20:24:29Z","links":{"resolver":"https://pith.science/pith/IE5AYRLKXPWXQAKDIA7VJZFBG6","bundle":"https://pith.science/pith/IE5AYRLKXPWXQAKDIA7VJZFBG6/bundle.json","state":"https://pith.science/pith/IE5AYRLKXPWXQAKDIA7VJZFBG6/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IE5AYRLKXPWXQAKDIA7VJZFBG6/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2021:IE5AYRLKXPWXQAKDIA7VJZFBG6","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":"e564a364b7e0fe99ed564846a62cc918cf01bf3be560237a4a1bbd759e1d6889","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-02T08:35:02Z","title_canon_sha256":"3c508e316ca55bbcfca33aa0a183f7e41673c16ccb37f5f748f2219bde0e26e0"},"schema_version":"1.0","source":{"id":"2102.01386","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2102.01386","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"arxiv_version","alias_value":"2102.01386v2","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.01386","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"pith_short_12","alias_value":"IE5AYRLKXPWX","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"pith_short_16","alias_value":"IE5AYRLKXPWXQAKD","created_at":"2026-07-05T02:28:55Z"},{"alias_kind":"pith_short_8","alias_value":"IE5AYRLK","created_at":"2026-07-05T02:28:55Z"}],"graph_snapshots":[{"event_id":"sha256:8f3126bb0aeb021679d5ddbc0af0532f9209d2faa44ee8643673cd7f93555c14","target":"graph","created_at":"2026-07-05T02:28:55Z","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/2102.01386/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"With the rapid adoption of machine learning (ML), a number of domains now use the approach of fine tuning models which were pre-trained on a large corpus of data. However, our experiments show that even fine-tuning on models like BERT can take many hours even when using modern accelerators like GPUs. While prior work proposes limiting the number of layers that are fine-tuned, e.g., freezing all layers but the last layer, we find that such static approaches lead to reduced accuracy. We propose, AutoFreeze, a system that uses an adaptive approach to choose which layers are trained and show how t","authors_text":"Saurabh Agarwal, Shivaram Venkataraman, Yuhan Liu","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-02T08:35:02Z","title":"AutoFreeze: Automatically Freezing Model Blocks to Accelerate Fine-tuning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.01386","kind":"arxiv","version":2},"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:d3363ecf58985fb10e662be084430ddd701a758fb878ab2e3660d422f7afc60f","target":"record","created_at":"2026-07-05T02:28:55Z","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":"e564a364b7e0fe99ed564846a62cc918cf01bf3be560237a4a1bbd759e1d6889","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-02T08:35:02Z","title_canon_sha256":"3c508e316ca55bbcfca33aa0a183f7e41673c16ccb37f5f748f2219bde0e26e0"},"schema_version":"1.0","source":{"id":"2102.01386","kind":"arxiv","version":2}},"canonical_sha256":"413a0c456abbed780143403f54e4a1378e2522d624a09193f8ace323fa792159","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"413a0c456abbed780143403f54e4a1378e2522d624a09193f8ace323fa792159","first_computed_at":"2026-07-05T02:28:55.466579Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T02:28:55.466579Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"idqrmvPREKIpXgJH/77KPAvs7PmTlw6mzT6CkGxEpCkr2k1Ka8WBQbj0AocZZPuXABcmxT/zMv/BVaGlKrw8Aw==","signature_status":"signed_v1","signed_at":"2026-07-05T02:28:55.467089Z","signed_message":"canonical_sha256_bytes"},"source_id":"2102.01386","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d3363ecf58985fb10e662be084430ddd701a758fb878ab2e3660d422f7afc60f","sha256:8f3126bb0aeb021679d5ddbc0af0532f9209d2faa44ee8643673cd7f93555c14"],"state_sha256":"bab56e95a4738f0a362126a8e66d5b8b4c412de739f318084846070327f8dbd9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vC6Swhh93LhcovSOblWerJdI2jC3ar75mYFfu5WAMVTe5g8R7OQ9rKAdFF33gZvOYnvTB1EpjJPLTfVsQf97Cg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T20:24:29.274227Z","bundle_sha256":"f8ab84c73e5db950ec7182454a788a76182fc0098624265f50b41d58dd951a07"}}