{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,25]],"date-time":"2026-03-25T06:21:30Z","timestamp":1774419690257,"version":"3.50.1"},"reference-count":38,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,4,6]],"date-time":"2025-04-06T00:00:00Z","timestamp":1743897600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,4,6]],"date-time":"2025-04-06T00:00:00Z","timestamp":1743897600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,4,6]]},"DOI":"10.1109\/icassp49660.2025.10887916","type":"proceedings-article","created":{"date-parts":[[2025,3,12]],"date-time":"2025-03-12T13:52:43Z","timestamp":1741787563000},"page":"1-5","source":"Crossref","is-referenced-by-count":0,"title":["CritiPrefill: A Segment-wise Criticality-based Approach for Prefilling Acceleration in LLMs"],"prefix":"10.1109","author":[{"given":"Junlin","family":"Lv","sequence":"first","affiliation":[{"name":"Guangdong OPPO Mobile Telecommunications Corp.,Ltd,AI Center,China"}]},{"given":"Yuan","family":"Feng","sequence":"additional","affiliation":[{"name":"Guangdong OPPO Mobile Telecommunications Corp.,Ltd,AI Center,China"}]},{"given":"Xike","family":"Xie","sequence":"additional","affiliation":[{"name":"Guangdong OPPO Mobile Telecommunications Corp.,Ltd,AI Center,China"}]},{"given":"Xin","family":"Jia","sequence":"additional","affiliation":[{"name":"Guangdong OPPO Mobile Telecommunications Corp.,Ltd,AI Center,China"}]},{"given":"Qirong","family":"Peng","sequence":"additional","affiliation":[{"name":"Guangdong OPPO Mobile Telecommunications Corp.,Ltd,AI Center,China"}]},{"given":"Guiming","family":"Xie","sequence":"additional","affiliation":[{"name":"Guangdong OPPO Mobile Telecommunications Corp.,Ltd,AI Center,China"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-main.600"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D18-1259"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-main.64"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1007\/s11704-024-40231-1"},{"key":"ref5","article-title":"Efficient large language models: A survey","author":"Wan","year":"2024"},{"key":"ref6","article-title":"A survey on long text modeling with transformers","author":"Dong","year":"2023"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.24963\/ijcai.2024\/634"},{"key":"ref8","article-title":"Linformer: Self-attention with linear complexity","author":"Wang","year":"2020"},{"key":"ref9","article-title":"Mooncake: A kvcache-centric disaggregated architecture for llm serving","author":"Qin","year":"2024"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.607"},{"key":"ref11","first-page":"16344","article-title":"Flashattention: Fast and memory-efficient exact attention with io-awareness","volume":"35","author":"Dao","year":"2022","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/3600006.3613165"},{"key":"ref13","article-title":"Swift: On-the-fly self-speculative decoding for llm inference acceleration","author":"Xia","year":"2024"},{"key":"ref14","article-title":"Ada-kv: Optimizing kv cache eviction by adaptive budget allocation for efficient llm inference","author":"Feng","year":"2024"},{"key":"ref15","article-title":"Snapkv: Llm knows what you are looking for before generation","author":"Li","year":"2024"},{"key":"ref16","article-title":"H2o: Heavy-hitter oracle for efficient generative inference of large language models","volume":"36","author":"Zhang","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref17","article-title":"Quest: Query-aware sparsity for efficient long-context llm inference","author":"Tang","year":"2024"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/wacv48630.2021.00357"},{"key":"ref19","article-title":"Gated linear attention transformers with hardware-efficient training","author":"Yang","year":"2023"},{"key":"ref20","article-title":"Longformer: The long-document transformer","author":"Beltagy","year":"2020"},{"key":"ref21","first-page":"17283","article-title":"Big bird: Transformers for longer sequences","volume":"33","author":"Zaheer","year":"2020","journal-title":"Advances in neural information processing systems"},{"key":"ref22","article-title":"Focused transformer: Contrastive training for context scaling","volume":"36","author":"Tworkowski","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1145\/3394486.3406703"},{"key":"ref24","article-title":"Megatron-lm: Training multi-billion parameter language models using model parallelism","author":"Shoeybi","year":"2019"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P19-1285"},{"key":"ref26","first-page":"22137","article-title":"Deja vu: Contextual sparsity for efficient llms at inference time","volume-title":"International Conference on Machine Learning","author":"Liu"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-emnlp.183"},{"key":"ref28","article-title":"Model tells you what to discard: Adaptive kv cache compression for llms","author":"Ge","year":"2023"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.681"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00023"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.naacl-main.365"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.172"},{"key":"ref33","article-title":"Lv-eval: A balanced long-context benchmark with 5 length levels up to 256k","author":"Yuan","year":"2024"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00638"},{"key":"ref35","article-title":"Llama 2: Open foundation and fine-tuned chat models","author":"Touvron","year":"2023"},{"key":"ref36","article-title":"Yi: Open foundation models by 01.ai","author":"Young","year":"2024"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.naacl-long.260"},{"key":"ref38","article-title":"Flashattention-2: Faster attention with better parallelism and work partitioning","author":"Dao","year":"2023"}],"event":{"name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","location":"Hyderabad, India","start":{"date-parts":[[2025,4,6]]},"end":{"date-parts":[[2025,4,11]]}},"container-title":["ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/10887540\/10887541\/10887916.pdf?arnumber=10887916","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,25]],"date-time":"2026-03-25T05:24:08Z","timestamp":1774416248000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10887916\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,4,6]]},"references-count":38,"URL":"https:\/\/doi.org\/10.1109\/icassp49660.2025.10887916","relation":{},"subject":[],"published":{"date-parts":[[2025,4,6]]}}}