{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KGEMQT4JVGZFELBV5U27TIDAGI","short_pith_number":"pith:KGEMQT4J","schema_version":"1.0","canonical_sha256":"5188c84f89a9b2522c35ed35f9a06032124aee67c501bf88de0abbf8fe92e0d6","source":{"kind":"arxiv","id":"2505.16983","version":2},"attestation_state":"computed","paper":{"title":"LLM as Effective Streaming Processor: Bridging Streaming-Batch Mismatches with Group Position Encoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anhao Zhao, Hui Su, Jinlan Fu, Junlong Tong, Xiaoyu Shen, Yingqi Fan, Zixuan Lin","submitted_at":"2025-05-22T17:53:28Z","abstract_excerpt":"Large Language Models (LLMs) are primarily designed for batch processing. Existing methods for adapting LLMs to streaming rely either on expensive re-encoding or specialized architectures with limited scalability. This work identifies three key mismatches in adapting batch-oriented LLMs to streaming: (1) input-attention, (2) output-attention, and (3) position-ID mismatches. While it is commonly assumed that the latter two mismatches require frequent re-encoding, our analysis reveals that only the input-attention mismatch significantly impacts performance, indicating re-encoding outputs is larg"},"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":"2505.16983","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-22T17:53:28Z","cross_cats_sorted":[],"title_canon_sha256":"cf260790e2d27c6b8899957e58f50ad80cb0a333a02b8d90ebc85ce8a9337526","abstract_canon_sha256":"25ab499954df3f0da91b8d8c18ae9faef030ad7ac11382549d09b0db2866872e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:11:46.060695Z","signature_b64":"IZObkUoYzwTb5eGQ1e2Kn6hcFoNxbv4vpXYV1vuObiSF31DCn5FWzhoV8OujZH6esYc/br2Arqz9QhuXJX9sCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5188c84f89a9b2522c35ed35f9a06032124aee67c501bf88de0abbf8fe92e0d6","last_reissued_at":"2026-07-05T11:11:46.060043Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:11:46.060043Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLM as Effective Streaming Processor: Bridging Streaming-Batch Mismatches with Group Position Encoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anhao Zhao, Hui Su, Jinlan Fu, Junlong Tong, Xiaoyu Shen, Yingqi Fan, Zixuan Lin","submitted_at":"2025-05-22T17:53:28Z","abstract_excerpt":"Large Language Models (LLMs) are primarily designed for batch processing. Existing methods for adapting LLMs to streaming rely either on expensive re-encoding or specialized architectures with limited scalability. This work identifies three key mismatches in adapting batch-oriented LLMs to streaming: (1) input-attention, (2) output-attention, and (3) position-ID mismatches. While it is commonly assumed that the latter two mismatches require frequent re-encoding, our analysis reveals that only the input-attention mismatch significantly impacts performance, indicating re-encoding outputs is larg"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.16983","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/2505.16983/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":"2505.16983","created_at":"2026-07-05T11:11:46.060117+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.16983v2","created_at":"2026-07-05T11:11:46.060117+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.16983","created_at":"2026-07-05T11:11:46.060117+00:00"},{"alias_kind":"pith_short_12","alias_value":"KGEMQT4JVGZF","created_at":"2026-07-05T11:11:46.060117+00:00"},{"alias_kind":"pith_short_16","alias_value":"KGEMQT4JVGZFELBV","created_at":"2026-07-05T11:11:46.060117+00:00"},{"alias_kind":"pith_short_8","alias_value":"KGEMQT4J","created_at":"2026-07-05T11:11:46.060117+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.00523","citing_title":"ProactiveLLM: Learning Active Interaction for Streaming Large Language Models","ref_index":99,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI","json":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI.json","graph_json":"https://pith.science/api/pith-number/KGEMQT4JVGZFELBV5U27TIDAGI/graph.json","events_json":"https://pith.science/api/pith-number/KGEMQT4JVGZFELBV5U27TIDAGI/events.json","paper":"https://pith.science/paper/KGEMQT4J"},"agent_actions":{"view_html":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI","download_json":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI.json","view_paper":"https://pith.science/paper/KGEMQT4J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.16983&json=true","fetch_graph":"https://pith.science/api/pith-number/KGEMQT4JVGZFELBV5U27TIDAGI/graph.json","fetch_events":"https://pith.science/api/pith-number/KGEMQT4JVGZFELBV5U27TIDAGI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI/action/storage_attestation","attest_author":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI/action/author_attestation","sign_citation":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI/action/citation_signature","submit_replication":"https://pith.science/pith/KGEMQT4JVGZFELBV5U27TIDAGI/action/replication_record"}},"created_at":"2026-07-05T11:11:46.060117+00:00","updated_at":"2026-07-05T11:11:46.060117+00:00"}