{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:FNGKPJ7WBQ6HYTYLIPVLX6Q7RP","short_pith_number":"pith:FNGKPJ7W","canonical_record":{"source":{"id":"2607.08284","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-09T09:28:06Z","cross_cats_sorted":[],"title_canon_sha256":"cdf6a5f896231aa93cdc1a77d2d9875883a13aca99b91825c33ae2d60ee75a46","abstract_canon_sha256":"9de5348211b1a7cf47310b6f63754486f93af3ff6956cb894a24c920b160d73b"},"schema_version":"1.0"},"canonical_sha256":"2b4ca7a7f60c3c7c4f0b43eabbfa1f8bde49619eb7c2c9c10a46a108973efb05","source":{"kind":"arxiv","id":"2607.08284","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.08284","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"arxiv_version","alias_value":"2607.08284v1","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.08284","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"pith_short_12","alias_value":"FNGKPJ7WBQ6H","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"pith_short_16","alias_value":"FNGKPJ7WBQ6HYTYL","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"pith_short_8","alias_value":"FNGKPJ7W","created_at":"2026-07-10T01:19:31Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:FNGKPJ7WBQ6HYTYLIPVLX6Q7RP","target":"record","payload":{"canonical_record":{"source":{"id":"2607.08284","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-09T09:28:06Z","cross_cats_sorted":[],"title_canon_sha256":"cdf6a5f896231aa93cdc1a77d2d9875883a13aca99b91825c33ae2d60ee75a46","abstract_canon_sha256":"9de5348211b1a7cf47310b6f63754486f93af3ff6956cb894a24c920b160d73b"},"schema_version":"1.0"},"canonical_sha256":"2b4ca7a7f60c3c7c4f0b43eabbfa1f8bde49619eb7c2c9c10a46a108973efb05","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-10T01:19:31.838092Z","signature_b64":"Xi+efg2/tJ7nYPjIoDVIFM7RLAAvIbJxRUr/voWsw6fb1JisshKOgzbSgoAyqwADbEysui/odiXtmXFFdRnPCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2b4ca7a7f60c3c7c4f0b43eabbfa1f8bde49619eb7c2c9c10a46a108973efb05","last_reissued_at":"2026-07-10T01:19:31.837721Z","signature_status":"signed_v1","first_computed_at":"2026-07-10T01:19:31.837721Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2607.08284","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-10T01:19:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0Tt51o1Vs+CtYEahPAGSGyqGHlB+9n7sSYi4v48h7vrsWQTHkOiPXEcXEdQTR30kAD413z2hv8ViPMjlUm3+DA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T09:13:24.508940Z"},"content_sha256":"935a01491cbb9895092476de65044c53b3ce7d805cffcf001d0101b1b6bbef0c","schema_version":"1.0","event_id":"sha256:935a01491cbb9895092476de65044c53b3ce7d805cffcf001d0101b1b6bbef0c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:FNGKPJ7WBQ6HYTYLIPVLX6Q7RP","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Understanding Axes of Difficulty For Long Context Tasks Via PredicateLongBench","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Ameya Velingker, Siddhartha Jain","submitted_at":"2026-07-09T09:28:06Z","abstract_excerpt":"Large language models (LLMs) have demonstrated rapidly improving long-context capabilities, prompting a wave of benchmarks designed to evaluate them. However, existing long-context evaluations - from Needle-in-a-Haystack (NIAH) tests to more recent multi-hop reasoning and summarization tasks - predominantly measure average-case performance, and many are either saturated or lack robustness. Notably absent is a systematic way to probe how models perform as we scale up the difficulty of tasks along various axes. We address this gap by proposing PredicateLongBench, a benchmark that stress-tests lo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.08284","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/2607.08284/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":"2026-07-10T01:19:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Y0PKd16KsSFlyOaA8sy62I3YPcGblYTekTYJ9yMeJGDRxWzVnmcuF8R11/da3g7dBKk7zy0S4LOeGzHbZbaHCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T09:13:24.509498Z"},"content_sha256":"6344897fc3b849f00ad3e7053ea3157ab1a0ab48e3a3144a7b751f7224a8df25","schema_version":"1.0","event_id":"sha256:6344897fc3b849f00ad3e7053ea3157ab1a0ab48e3a3144a7b751f7224a8df25"},{"event_type":"integrity_finding","subject_pith_number":"pith:2026:FNGKPJ7WBQ6HYTYLIPVLX6Q7RP","target":"integrity","payload":{"note":"DOI in the printed bibliography is fragmented by whitespace or line breaks. A longer candidate (10.48550/arXiv.2406) was visible in the surrounding text but could not be confirmed against doi.org as printed.","snippet":"doi: 10.48550/ARXIV .2406.06025. URL https://doi.org/10.48550/arXiv.2406. 06025. Ali Modarressi, Hanieh Deilamsalehy, Franck Dernoncourt, Trung Bui, Ryan A. Rossi, Seunghyun Yoon, and Hinrich Schütze. Nolima: Long-context evaluation beyond ","arxiv_id":"2607.08284","detector":"doi_compliance","evidence":{"ref_index":2,"verdict_class":"incontrovertible","resolved_title":null,"printed_excerpt":"doi: 10.48550/ARXIV .2406.06025. URL https://doi.org/10.48550/arXiv.2406. 06025. Ali Modarressi, Hanieh Deilamsalehy, Franck Dernoncourt, Trung Bui, Ryan A. Rossi, Seunghyun Yoon, and Hinrich Schütze. Nolima: Long-context evaluation beyond ","reconstructed_doi":"10.48550/arXiv.2406"},"severity":"advisory","ref_index":2,"audited_at":"2026-07-10T10:12:48.980900Z","event_type":"pith.integrity.v1","detected_doi":"10.48550/arXiv.2406","detector_url":"https://pith.science/pith-integrity-protocol#doi_compliance","external_url":null,"finding_type":"recoverable_identifier","evidence_hash":"d3f4de166c9cc64027fe40a46de39d40f5c7eb2561a795d76b31e40ca8402b0f","paper_version":1,"verdict_class":"incontrovertible","resolved_title":null,"detector_version":"1.0.0","detected_arxiv_id":null,"integrity_event_id":11926,"payload_sha256":"341420670439400224a0741e02c0105a22a8ddf0eee5dd0c73387ca17aa0cb08","signature_b64":"W4lRNBilhyrZ8fAPH8GHe6EJ8lOySC/tfvWF8mXYgGM9GYh3V+MoO3HdbjjD/ydsrEZU1Z9/CLkQp04H8Q6KAg==","signing_key_id":"pith-v1-2026-05"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-10T10:14:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"62+bFPYEplAjmLdQ25X700AaTF9FFwFUCMcxyED6Rb3jcGQxtCPKcZsDn+TStEytQjoMqtcAFgAkHtlZahfoBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T09:13:24.511380Z"},"content_sha256":"821a60dbd3df607241faf01893202d808088c58f905673aa9bae2a3ec45aea6b","schema_version":"1.0","event_id":"sha256:821a60dbd3df607241faf01893202d808088c58f905673aa9bae2a3ec45aea6b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FNGKPJ7WBQ6HYTYLIPVLX6Q7RP/bundle.json","state_url":"https://pith.science/pith/FNGKPJ7WBQ6HYTYLIPVLX6Q7RP/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FNGKPJ7WBQ6HYTYLIPVLX6Q7RP/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":"2026-08-05T09:13:24Z","links":{"resolver":"https://pith.science/pith/FNGKPJ7WBQ6HYTYLIPVLX6Q7RP","bundle":"https://pith.science/pith/FNGKPJ7WBQ6HYTYLIPVLX6Q7RP/bundle.json","state":"https://pith.science/pith/FNGKPJ7WBQ6HYTYLIPVLX6Q7RP/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FNGKPJ7WBQ6HYTYLIPVLX6Q7RP/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:FNGKPJ7WBQ6HYTYLIPVLX6Q7RP","merge_version":"pith-open-graph-merge-v1","event_count":3,"valid_event_count":3,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"9de5348211b1a7cf47310b6f63754486f93af3ff6956cb894a24c920b160d73b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-09T09:28:06Z","title_canon_sha256":"cdf6a5f896231aa93cdc1a77d2d9875883a13aca99b91825c33ae2d60ee75a46"},"schema_version":"1.0","source":{"id":"2607.08284","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.08284","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"arxiv_version","alias_value":"2607.08284v1","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.08284","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"pith_short_12","alias_value":"FNGKPJ7WBQ6H","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"pith_short_16","alias_value":"FNGKPJ7WBQ6HYTYL","created_at":"2026-07-10T01:19:31Z"},{"alias_kind":"pith_short_8","alias_value":"FNGKPJ7W","created_at":"2026-07-10T01:19:31Z"}],"graph_snapshots":[{"event_id":"sha256:6344897fc3b849f00ad3e7053ea3157ab1a0ab48e3a3144a7b751f7224a8df25","target":"graph","created_at":"2026-07-10T01:19:31Z","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/2607.08284/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) have demonstrated rapidly improving long-context capabilities, prompting a wave of benchmarks designed to evaluate them. However, existing long-context evaluations - from Needle-in-a-Haystack (NIAH) tests to more recent multi-hop reasoning and summarization tasks - predominantly measure average-case performance, and many are either saturated or lack robustness. Notably absent is a systematic way to probe how models perform as we scale up the difficulty of tasks along various axes. We address this gap by proposing PredicateLongBench, a benchmark that stress-tests lo","authors_text":"Ameya Velingker, Siddhartha Jain","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-09T09:28:06Z","title":"Understanding Axes of Difficulty For Long Context Tasks Via PredicateLongBench"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.08284","kind":"arxiv","version":1},"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":"sha256:935a01491cbb9895092476de65044c53b3ce7d805cffcf001d0101b1b6bbef0c","target":"record","created_at":"2026-07-10T01:19:31Z","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":"9de5348211b1a7cf47310b6f63754486f93af3ff6956cb894a24c920b160d73b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-09T09:28:06Z","title_canon_sha256":"cdf6a5f896231aa93cdc1a77d2d9875883a13aca99b91825c33ae2d60ee75a46"},"schema_version":"1.0","source":{"id":"2607.08284","kind":"arxiv","version":1}},"canonical_sha256":"2b4ca7a7f60c3c7c4f0b43eabbfa1f8bde49619eb7c2c9c10a46a108973efb05","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2b4ca7a7f60c3c7c4f0b43eabbfa1f8bde49619eb7c2c9c10a46a108973efb05","first_computed_at":"2026-07-10T01:19:31.837721Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-10T01:19:31.837721Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Xi+efg2/tJ7nYPjIoDVIFM7RLAAvIbJxRUr/voWsw6fb1JisshKOgzbSgoAyqwADbEysui/odiXtmXFFdRnPCg==","signature_status":"signed_v1","signed_at":"2026-07-10T01:19:31.838092Z","signed_message":"canonical_sha256_bytes"},"source_id":"2607.08284","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:935a01491cbb9895092476de65044c53b3ce7d805cffcf001d0101b1b6bbef0c","sha256:6344897fc3b849f00ad3e7053ea3157ab1a0ab48e3a3144a7b751f7224a8df25","sha256:821a60dbd3df607241faf01893202d808088c58f905673aa9bae2a3ec45aea6b"],"state_sha256":"ba5dc60b5b8a5edb73392cafd600c373227fecaf9226eea3b2b73d60165ddd8d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6O3AZF2GYcaQUTCO8QAnrMePPhYJo8pTlAOHu0/mBkrlzX/vgyZhCbu8pd/yHEvybvfsufoaRwLEu5tG6ix6Ag==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T09:13:24.513397Z","bundle_sha256":"408eab97360048f81f1441b57c5128749ea7d2a26aa0cee6e60ca11f754296de"}}