{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:SN2NVYBID5RMCFLPP6FK7ZHW7J","short_pith_number":"pith:SN2NVYBI","canonical_record":{"source":{"id":"2607.00448","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.IR","submitted_at":"2026-07-01T05:07:32Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"784ea7f95355e05779b45b4bac0ae407006609d8cc37042361979c8f7fcca41f","abstract_canon_sha256":"fb8ba78ca86a69badcdc928aec065c31eb30c3c167b6a69cc4d45bd0ed4db602"},"schema_version":"1.0"},"canonical_sha256":"9374dae0281f62c1156f7f8aafe4f6fa5cbff00b2191c7e4300758f9cc3a33c9","source":{"kind":"arxiv","id":"2607.00448","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.00448","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"arxiv_version","alias_value":"2607.00448v1","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.00448","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"pith_short_12","alias_value":"SN2NVYBID5RM","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"pith_short_16","alias_value":"SN2NVYBID5RMCFLP","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"pith_short_8","alias_value":"SN2NVYBI","created_at":"2026-07-02T01:17:43Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:SN2NVYBID5RMCFLPP6FK7ZHW7J","target":"record","payload":{"canonical_record":{"source":{"id":"2607.00448","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.IR","submitted_at":"2026-07-01T05:07:32Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"784ea7f95355e05779b45b4bac0ae407006609d8cc37042361979c8f7fcca41f","abstract_canon_sha256":"fb8ba78ca86a69badcdc928aec065c31eb30c3c167b6a69cc4d45bd0ed4db602"},"schema_version":"1.0"},"canonical_sha256":"9374dae0281f62c1156f7f8aafe4f6fa5cbff00b2191c7e4300758f9cc3a33c9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-02T01:17:43.680816Z","signature_b64":"fzFcNu99u1qQFgqBjVguRpyFZrCtss85Akm2ZO/akpw83rNgnLyON6Ur8DmqW+Df9xUdmGhsFXDqXIAEO1XdAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9374dae0281f62c1156f7f8aafe4f6fa5cbff00b2191c7e4300758f9cc3a33c9","last_reissued_at":"2026-07-02T01:17:43.680394Z","signature_status":"signed_v1","first_computed_at":"2026-07-02T01:17:43.680394Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2607.00448","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":"2026-07-02T01:17:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vaM56Sp59hk+H/OY3jFYhmUsmBm3gD/V8KCt0JPBGl6BA3OBO9f3jJesnk8r2pev0oBzdPv96Cg4QZL5K2ezBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-03T21:54:15.194174Z"},"content_sha256":"0600f299c72dc238013a3a8d247eb5788f27031faec540b67fab87ebab463f61","schema_version":"1.0","event_id":"sha256:0600f299c72dc238013a3a8d247eb5788f27031faec540b67fab87ebab463f61"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:SN2NVYBID5RMCFLPP6FK7ZHW7J","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Real-Time Hard Negative Sampling via LLM-based Clustering for Large-Scale Two-Tower Retrieval","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.IR","authors_text":"Aameek Singh, Harrison (Zihao) Zhao, Ivan Ji, Lei Huang, Liuyi Hu, Max (Xiangjun) Fan, Qunshu Zhang","submitted_at":"2026-07-01T05:07:32Z","abstract_excerpt":"The two-tower model has been widely used for large-scale recommendation systems, particularly in the retrieval stage. Industry standards for training two-tower models typically involve in-batch and/or out-of-batch negative sampling. However, these methods often produce easy negatives that models can quickly learn, failing to sufficiently challenge the model. To address this issue, a novel self-supervised hard negative sampling technique is proposed that leverages a large language model (LLM) to generate hard negatives from the same cluster during model training. By utilizing the LLM to learn m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.00448","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/2607.00448/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":"2026-07-02T01:17:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GobfbHJ62rL1T/H/DET+tlqGbPpQT0emY2IsjCXQyGuHw/T0F6sW2tlwtFUdrx9qpmHvrB3wFkdYy0OBQod0Bw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-03T21:54:15.194826Z"},"content_sha256":"c59d9736b528f11c030e440895a8c27bacaaee79f2d0f39b1e64d3727c303b04","schema_version":"1.0","event_id":"sha256:c59d9736b528f11c030e440895a8c27bacaaee79f2d0f39b1e64d3727c303b04"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SN2NVYBID5RMCFLPP6FK7ZHW7J/bundle.json","state_url":"https://pith.science/pith/SN2NVYBID5RMCFLPP6FK7ZHW7J/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SN2NVYBID5RMCFLPP6FK7ZHW7J/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":"2026-07-03T21:54:15Z","links":{"resolver":"https://pith.science/pith/SN2NVYBID5RMCFLPP6FK7ZHW7J","bundle":"https://pith.science/pith/SN2NVYBID5RMCFLPP6FK7ZHW7J/bundle.json","state":"https://pith.science/pith/SN2NVYBID5RMCFLPP6FK7ZHW7J/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SN2NVYBID5RMCFLPP6FK7ZHW7J/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:SN2NVYBID5RMCFLPP6FK7ZHW7J","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":"fb8ba78ca86a69badcdc928aec065c31eb30c3c167b6a69cc4d45bd0ed4db602","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.IR","submitted_at":"2026-07-01T05:07:32Z","title_canon_sha256":"784ea7f95355e05779b45b4bac0ae407006609d8cc37042361979c8f7fcca41f"},"schema_version":"1.0","source":{"id":"2607.00448","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.00448","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"arxiv_version","alias_value":"2607.00448v1","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.00448","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"pith_short_12","alias_value":"SN2NVYBID5RM","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"pith_short_16","alias_value":"SN2NVYBID5RMCFLP","created_at":"2026-07-02T01:17:43Z"},{"alias_kind":"pith_short_8","alias_value":"SN2NVYBI","created_at":"2026-07-02T01:17:43Z"}],"graph_snapshots":[{"event_id":"sha256:c59d9736b528f11c030e440895a8c27bacaaee79f2d0f39b1e64d3727c303b04","target":"graph","created_at":"2026-07-02T01:17:43Z","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/2607.00448/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The two-tower model has been widely used for large-scale recommendation systems, particularly in the retrieval stage. Industry standards for training two-tower models typically involve in-batch and/or out-of-batch negative sampling. However, these methods often produce easy negatives that models can quickly learn, failing to sufficiently challenge the model. To address this issue, a novel self-supervised hard negative sampling technique is proposed that leverages a large language model (LLM) to generate hard negatives from the same cluster during model training. By utilizing the LLM to learn m","authors_text":"Aameek Singh, Harrison (Zihao) Zhao, Ivan Ji, Lei Huang, Liuyi Hu, Max (Xiangjun) Fan, Qunshu Zhang","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.IR","submitted_at":"2026-07-01T05:07:32Z","title":"Real-Time Hard Negative Sampling via LLM-based Clustering for Large-Scale Two-Tower Retrieval"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.00448","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":"sha256:0600f299c72dc238013a3a8d247eb5788f27031faec540b67fab87ebab463f61","target":"record","created_at":"2026-07-02T01:17:43Z","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":"fb8ba78ca86a69badcdc928aec065c31eb30c3c167b6a69cc4d45bd0ed4db602","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.IR","submitted_at":"2026-07-01T05:07:32Z","title_canon_sha256":"784ea7f95355e05779b45b4bac0ae407006609d8cc37042361979c8f7fcca41f"},"schema_version":"1.0","source":{"id":"2607.00448","kind":"arxiv","version":1}},"canonical_sha256":"9374dae0281f62c1156f7f8aafe4f6fa5cbff00b2191c7e4300758f9cc3a33c9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9374dae0281f62c1156f7f8aafe4f6fa5cbff00b2191c7e4300758f9cc3a33c9","first_computed_at":"2026-07-02T01:17:43.680394Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-02T01:17:43.680394Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"fzFcNu99u1qQFgqBjVguRpyFZrCtss85Akm2ZO/akpw83rNgnLyON6Ur8DmqW+Df9xUdmGhsFXDqXIAEO1XdAQ==","signature_status":"signed_v1","signed_at":"2026-07-02T01:17:43.680816Z","signed_message":"canonical_sha256_bytes"},"source_id":"2607.00448","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0600f299c72dc238013a3a8d247eb5788f27031faec540b67fab87ebab463f61","sha256:c59d9736b528f11c030e440895a8c27bacaaee79f2d0f39b1e64d3727c303b04"],"state_sha256":"d9989677ec52f6e6ce15416c1c43eac05a68446e379bd5efb6a9ed444eed56d9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"W65j4WnAusmoqWBRQaJjX5B3NF2ldMUtmPyBwRLc/nqSgRuELo7BHvduZpmTvlv4o3aKRXv3/Z1UEqHB7Mr5Bw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-03T21:54:15.197807Z","bundle_sha256":"8749616ddff8881dcb324193da729f722e97b4020274860bafd03f78e52a061b"}}