{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:4HT3Q46XOBKAPZZHJBA2ZOWJZ4","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":"bd29f8c25355fee4f2d5f38a0228802a52c9dc460029c06a1cb59b935799d4dc","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T17:06:15Z","title_canon_sha256":"c02974716d1ef298f9e1780a9054c448f6d79e4fd29987f4ce8faa7df932c10d"},"schema_version":"1.0","source":{"id":"2605.13784","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13784","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13784v1","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13784","created_at":"2026-05-18T02:44:15Z"},{"alias_kind":"pith_short_12","alias_value":"4HT3Q46XOBKA","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"4HT3Q46XOBKAPZZH","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"4HT3Q46X","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:5eacd447a71483709360535c6a18cabfe505a9f00dfafddf67c044926eb8f901","target":"graph","created_at":"2026-05-18T02:44:15Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"On streaming market-data benchmarks the reference implementation achieves up to 5.9x speedup over conventional inference engines (vLLM, SGLang, TensorRT-LLM, llama.cpp), holding query latency constant as accumulated context grows."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That a multi-tenant continuous-batching scheduler with cell-budget admission and prefix-aware grouped prefill can maintain full quadratic self-attention across dozens of stateful sessions without correctness loss or prohibitive overhead."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Stateful sessions with incremental KV cache and flash queries allow O(|q|) latency in streaming transformer inference, delivering up to 5.9x speedup over conventional engines while preserving full attention."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Stateful sessions with persistent KV caches let streaming transformer queries run in time independent of context size."}],"snapshot_sha256":"207dd651ac10f9a3c9b729cc7b456f8ba4fe592ec44a60c35d50ea43a45c49ba"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Conventional transformer inference engines are request-driven, paying an O(n) prefill cost on every query. In streaming workloads, where data arrives continuously and queries probe an ever-growing context, this cost is prohibitive. We introduce a data-driven computational model centred on stateful sessions: a persistent KV cache advanced incrementally as new data arrives, so prefill is moved off the critical path and query latency becomes O(|q|), independent of accumulated context size. Building on this, Flash Queries reclaim idle GPU cycles between data arrivals to pre-evaluate registered que","authors_text":"Victor Norgren","cross_cats":[],"headline":"Stateful sessions with persistent KV caches let streaming transformer queries run in time independent of context size.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T17:06:15Z","title":"Attention Once Is All You Need: Efficient Streaming Inference with Stateful Transformers"},"references":{"count":20,"internal_anchors":8,"resolved_work":20,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Lopez-Lira, A. and Tang, Y. Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models. SSRN, 2023","work_id":"7e8454c1-04ca-4350-8143-7510902ea219","year":2023},{"cited_arxiv_id":"2303.17564","doi":"","is_internal_anchor":true,"ref_index":2,"title":"BloombergGPT: A Large Language Model for Finance","work_id":"04e9adc6-bece-4ff4-a0f6-e836d88414a7","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Lost in the Middle: How Language Models Use Long Contexts","work_id":"1616dd1e-4de8-4717-a816-b131cad65926","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Prompt Caching","work_id":"278a7fa4-e0b4-4ec6-9bc3-fe1ff8995197","year":2024},{"cited_arxiv_id":"2004.05150","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Longformer: The Long-Document Transformer","work_id":"abea7a44-6668-4de7-aab6-f53a6e5aa088","year":2004}],"snapshot_sha256":"75e2bae35ff0b2d4f867ea960d54ddc78285ef194f5d5e5f29272cacaff5755d"},"source":{"id":"2605.13784","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T19:16:51.175879Z","id":"381654ce-1b87-4538-bc9b-f0672729f336","model_set":{"reader":"grok-4.3"},"one_line_summary":"Stateful sessions with incremental KV cache and flash queries allow O(|q|) latency in streaming transformer inference, delivering up to 5.9x speedup over conventional engines while preserving full attention.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Stateful sessions with persistent KV caches let streaming transformer queries run in time independent of context size.","strongest_claim":"On streaming market-data benchmarks the reference implementation achieves up to 5.9x speedup over conventional inference engines (vLLM, SGLang, TensorRT-LLM, llama.cpp), holding query latency constant as accumulated context grows.","weakest_assumption":"That a multi-tenant continuous-batching scheduler with cell-budget admission and prefix-aware grouped prefill can maintain full quadratic self-attention across dozens of stateful sessions without correctness loss or prohibitive overhead."}},"verdict_id":"381654ce-1b87-4538-bc9b-f0672729f336"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:23c8e9c676ee6650abbe051e211bd6c2a0fd43ddb6b1a8bdfdd3c5aade849adf","target":"record","created_at":"2026-05-18T02:44:15Z","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":"bd29f8c25355fee4f2d5f38a0228802a52c9dc460029c06a1cb59b935799d4dc","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T17:06:15Z","title_canon_sha256":"c02974716d1ef298f9e1780a9054c448f6d79e4fd29987f4ce8faa7df932c10d"},"schema_version":"1.0","source":{"id":"2605.13784","kind":"arxiv","version":1}},"canonical_sha256":"e1e7b873d7705407e7274841acbac9cf35dc7ad4a60bb028b6753d9d4fd70f87","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e1e7b873d7705407e7274841acbac9cf35dc7ad4a60bb028b6753d9d4fd70f87","first_computed_at":"2026-05-18T02:44:15.703620Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T02:44:15.703620Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"L9BxDxHKHnNaS6pG2CGPcogPB6Mgk/97YB9ezsg0lEE3FPmTIWykrzS20tEIYXZWsF2zpRoCFqplBaSAUVSJDw==","signature_status":"signed_v1","signed_at":"2026-05-18T02:44:15.704191Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.13784","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:23c8e9c676ee6650abbe051e211bd6c2a0fd43ddb6b1a8bdfdd3c5aade849adf","sha256:5eacd447a71483709360535c6a18cabfe505a9f00dfafddf67c044926eb8f901"],"state_sha256":"445ec1ca8f9d7406521e201e41fd2961045f7e0a4250e11af8c8cc3106253356"}