{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:NVAFL2CYTUBAUAGXMEJWSSWGB6","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":"ef8f3da93594914f409eba5625144ecd53a2459c3616175e48fa0c82a2e4b035","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2025-01-02T02:02:20Z","title_canon_sha256":"fec089e8cc02be9f63d92004ce4011f481ba5effd987267046f165313ae1587b"},"schema_version":"1.0","source":{"id":"2501.01005","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.01005","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"arxiv_version","alias_value":"2501.01005v2","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.01005","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"pith_short_12","alias_value":"NVAFL2CYTUBA","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"NVAFL2CYTUBAUAGX","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"NVAFL2CY","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:b695ebc832e179e785eb1464206db733a017a7d82ae6d04a317b5a0809d1d729","target":"graph","created_at":"2026-05-17T23:38:47Z","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":"Compared to state-of-the-art LLM serving solutions, FlashInfer achieve 29-69% inter-token-latency reduction compared to compiler backends for LLM serving benchmark, 28-30% latency reduction for long-context inference, and 13-17% speedup for LLM serving with parallel generation."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The reported speedups assume that the block-sparse format and JIT templates integrate cleanly with existing serving frameworks without hidden overheads from compilation or scheduling that would appear under production load patterns not tested in the benchmarks."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"FlashInfer delivers a customizable attention kernel that reduces inter-token latency by 29-69% in LLM serving benchmarks via optimized KV-cache storage and load-balanced scheduling compatible with CUDA graphs."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"FlashInfer uses block-sparse KV-cache formats and JIT-compiled attention templates to cut inter-token latency by 29-69% in LLM serving."}],"snapshot_sha256":"b4b44bab60ff148d0db56ba6d96d2b1a261dac13863d61f6f6848b3b77d3d25b"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"1d4b34b6a74bdcbf05457dc3fcba8230a664874232639143fd07d4e35971cce2"},"paper":{"abstract_excerpt":"Transformers, driven by attention mechanisms, form the foundation of large language models (LLMs). As these models scale up, efficient GPU attention kernels become essential for high-throughput and low-latency inference. Diverse LLM applications demand flexible and high-performance attention solutions. We present FlashInfer: a customizable and efficient attention engine for LLM serving. FlashInfer tackles KV-cache storage heterogeneity using block-sparse format and composable formats to optimize memory access and reduce redundancy. It also offers a customizable attention template, enabling ada","authors_text":"Arvind Krishnamurthy, Baris Kasikci, Lequn Chen, Luis Ceze, Ruihang Lai, Stephanie Wang, Tianqi Chen, Vinod Grover, Wuwei Lin, Yineng Zhang, Zihao Ye","cross_cats":["cs.AI","cs.LG"],"headline":"FlashInfer uses block-sparse KV-cache formats and JIT-compiled attention templates to cut inter-token latency by 29-69% in LLM serving.","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2025-01-02T02:02:20Z","title":"FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Serving"},"references":{"count":14,"internal_anchors":3,"resolved_work":14,"sample":[{"cited_arxiv_id":"","doi":"10.1145/1583991.1584053","is_internal_anchor":false,"ref_index":1,"title":"URL https://arxiv.org/abs/2004. 05150. Buluç, A., Fineman, J. T., Frigo, M., Gilbert, J. R., and Leiserson, C. E. Parallel sparse matrix-vector and matrix-transpose-vector multiplication using com- pr","work_id":"4e1afef5-52c1-4ec8-8c63-7c4bf85ce428","year":2004},{"cited_arxiv_id":"","doi":"10.1145/3458817.3476182","is_internal_anchor":false,"ref_index":2,"title":"URL https: //doi.org/10.1145/3458817.3476182","work_id":"f104f07e-0962-463f-959e-ddecd7f1f7ba","year":2024},{"cited_arxiv_id":"","doi":"10.1177/1094342004041296","is_internal_anchor":false,"ref_index":5,"title":"org/paper_files/paper/2024/file/ 5321b1dabcd2be188d796c21b733e8c7- Paper-Conference.pdf","work_id":"db04265b-d798-47c9-b726-4ca1021628a1","year":2024},{"cited_arxiv_id":"1404.2022","doi":"10.1109/sc41404.2022.00042","is_internal_anchor":true,"ref_index":7,"title":"El-rec: Efficient large-scale recommendation model training via tensor-train embedding table","work_id":"0625ddd0-436d-4c7a-b66a-97409cfc830c","year":2022},{"cited_arxiv_id":"","doi":"10.1145/55364.55378","is_internal_anchor":false,"ref_index":9,"title":"URL https://doi","work_id":"a5dd3d03-6672-4aae-acf0-c19739d4a596","year":2024}],"snapshot_sha256":"d5851aafbceae1d6888f616320a215806f2f3d523cfea2ed52d3d251580d9b49"},"source":{"id":"2501.01005","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-16T13:21:57.766174Z","id":"d2941e40-5d8c-4a82-b667-2b1836ce3227","model_set":{"reader":"grok-4.3"},"one_line_summary":"FlashInfer delivers a customizable attention kernel that reduces inter-token latency by 29-69% in LLM serving benchmarks via optimized KV-cache storage and load-balanced scheduling compatible with CUDA graphs.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"FlashInfer uses block-sparse KV-cache formats and JIT-compiled attention templates to cut inter-token latency by 29-69% in LLM serving.","strongest_claim":"Compared to state-of-the-art LLM serving solutions, FlashInfer achieve 29-69% inter-token-latency reduction compared to compiler backends for LLM serving benchmark, 28-30% latency reduction for long-context inference, and 13-17% speedup for LLM serving with parallel generation.","weakest_assumption":"The reported speedups assume that the block-sparse format and JIT templates integrate cleanly with existing serving frameworks without hidden overheads from compilation or scheduling that would appear under production load patterns not tested in the benchmarks."}},"verdict_id":"d2941e40-5d8c-4a82-b667-2b1836ce3227"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:fcb959c24dabb2f56dbddac69e3ead6457c096537af13e194ceb0a648660e4b0","target":"record","created_at":"2026-05-17T23:38:47Z","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":"ef8f3da93594914f409eba5625144ecd53a2459c3616175e48fa0c82a2e4b035","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2025-01-02T02:02:20Z","title_canon_sha256":"fec089e8cc02be9f63d92004ce4011f481ba5effd987267046f165313ae1587b"},"schema_version":"1.0","source":{"id":"2501.01005","kind":"arxiv","version":2}},"canonical_sha256":"6d4055e8589d020a00d76113694ac60fa3a2eb5e7fd631754622ac45fd3f6da5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6d4055e8589d020a00d76113694ac60fa3a2eb5e7fd631754622ac45fd3f6da5","first_computed_at":"2026-05-17T23:38:47.743252Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:47.743252Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"QoOhFvdckfGjafdkFh8Z9THlZcJhljkl1ltoAl1fR1BmUYV/jytGUrH7MDiOXxtr5DkZXQfjEQoN8lI/bQIJCQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:47.743715Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.01005","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:fcb959c24dabb2f56dbddac69e3ead6457c096537af13e194ceb0a648660e4b0","sha256:b695ebc832e179e785eb1464206db733a017a7d82ae6d04a317b5a0809d1d729"],"state_sha256":"00ff2b8b428787557fdd3ec05d3f9a79baec926e3fc01e519cbd7a035a611da4"}