Datasets:

fineset-io
/

long-context-llm-papers

id string	sources list	title string	abstract string	authors list	categories list	fields_of_study list	published_date timestamp[s]	url string	pdf_url string	arxiv_id string	doi string	citation_count int64	influential_citation_count int64	has_code bool	code_url string	venue string	quality_score float64
e404291ad74bdaaf1136be51797ed0d76ebcfda2dd2cfc13fba885c1abdcd582	[ "arxiv" ]	MemTrace: Probing What Final Accuracy Misses in Long-Term Memory	LLM agents increasingly maintain long-term memory of user facts across sessions. Yet such memory is usually evaluated by aggregating accuracy over question rows or episodes. Because this approach scores question rows independently, even when several questions probe the same fact, it cannot show how that fact behaves as...	[ "Xianxuan Long", "Zhikai Chen", "Shenglai Zeng", "Shouren Wang", "Kai Guo", "Jiliang Tang" ]	[ "cs.AI" ]	[]	2026-06-15T00:00:00	https://arxiv.org/abs/2606.17328	https://arxiv.org/pdf/2606.17328v1	2606.17328	null	0	0	false	null	null	0.35
1c35f195ee2726a4b934183c9eb8a1bc7960d6d30843e1eed8a99edbfdc0b1d1	[ "arxiv", "semantic_scholar" ]	SERF: Spatiotemporal Environment and Robot Feature Map for Long-Horizon Mobile Manipulation	Long-horizon robot mobile manipulation requires continual reasoning about localization, environment changes, and task progress, all of which are challenging to infer from image observations alone. In this paper, we show that conditioning a mobile manipulation policy on a spatiotemporal feature map improves reasoning ov...	[ "Sunghwan Kim", "Byeonghyun Pak", "Kehan Long", "Yulun Tian", "Nikolay Atanasov" ]	[ "cs.RO" ]	[ "Computer Science" ]	2026-06-11T00:00:00	https://arxiv.org/abs/2606.12956	https://arxiv.org/pdf/2606.12956v1	2606.12956	null	0	0	false	null	null	0.35
0e59872f2a812c9afa077800b0f4f6d236ee0f893871a478b149a89d70a9f1db	[ "arxiv", "semantic_scholar" ]	Unified KV Pooling to Accelerate Long-Context LLM Serving	Long-context LLM serving requires offloading KV caches to host-memory and SSDs, but existing mechanisms are not designed for such long contexts. We observe significant inefficiencies in current KV caching in long contexts: high serving latency ~30.7 s, exceeding the typical TTFT requirement of 10 s by more than 3x. Our...	[ "Minchul Kang", "Changyong Shin", "Jinwoo Jeong", "Jaerim Park", "Woohyun Kim", "Bonyul Gu", "Dongwoo Kang", "Gyeongsik Yang", "Chuck Yoo" ]	[ "cs.AR" ]	[ "Computer Science" ]	2026-06-10T00:00:00	https://arxiv.org/abs/2606.14779	https://arxiv.org/pdf/2606.14779v1	2606.14779	null	0	0	false	null	null	0.35
f237f3032c183164f6e9ef474274e99437fb3d91179fb5ced7d13f71a233a5eb	[ "arxiv", "semantic_scholar" ]	AHA-WAM:Asynchronous Horizon-Adaptive World-Action Modeling with Observation-Guided Context Routing	World-action models have emerged as a promising paradigm for robot manipulation, jointly modeling visual scene dynamics and actions to inject physical priors into policy learning. However, existing world-action models couple world prediction and action execution at the same temporal resolution, forcing the world branch...	[ "Jisong Cai", "Long Ling", "Shiwei Chu", "Zhongshan Liu", "Jiayue Kang", "Zhixuan Liang", "Wenjie Xu", "Yinan Mao", "Weinan Zhang", "Xiaokang Yang", "Ru Ying", "Ran Zheng", "Yao Mu" ]	[ "cs.RO", "cs.AI", "cs.CV" ]	[ "Computer Science" ]	2026-06-08T00:00:00	https://arxiv.org/abs/2606.09811	https://arxiv.org/pdf/2606.09811v1	2606.09811	null	0	0	false	null	null	0.35
5ea3182923a447700ff7de9a75dc9d912a92dbc16a5d4fdefccb458d42d9e167	[ "arxiv", "semantic_scholar" ]	YouZhi: Towards High-Concurrency Financial LLMs via Adaptive GQA-to-MLA Transition	Large language models (LLMs) drive significant financial innovations, yet their high-concurrency deployment is severely bottlenecked by KV cache memory overhead, which inflates infrastructure costs and throttles scalability. To address this, we propose YouZhi-LLM, a highly efficient financial LLM empowered by a compreh...	[ " PSBC LLM Team", " Huawei LLM Team", "Ruihan Long", "Junjie Wu", "Tianan Zhang", "Duo Zhang", "Yaozong Wu", "Jinbin Fu", "Chang Liu", "Zhentao Tang", "Wenshuang Yang", "Xin Wang", "Zhihao Song", "Ning Huang", "Wenjing Xu", "Shuai Zong", "Shupei Sun", "Sen Wang", "Jing Hu", "Bi...	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.05868	https://arxiv.org/pdf/2606.05868v1	2606.05868	null	0	0	false	null	null	0.35
a1e2b526ef175f3378540718c9054bdf942b9e4bd6c77f12afb45ee7a5a5fb84	[ "arxiv", "semantic_scholar" ]	Dense Contexts Are Hard Contexts: Lexical Density Limits Effective Context in LLMs	Input length and the position of relevant information are widely cited as the primary causes of degraded LLM long-context performance. Here, we study lexical density -- the rate at which a context introduces distinct information -- as a third, largely overlooked factor that systematically reduces the effective context ...	[ "Giovanni Dettori", "Matteo Boffa", "Danilo Giordano", "Idilio Drago", "Marco Mellia" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.06203	https://arxiv.org/pdf/2606.06203v1	2606.06203	null	0	0	false	null	null	0.35
b17032356cad3b28cfd5ee622d61801a67102ec0cddc842acccc425f30c6d0a4	[ "arxiv", "semantic_scholar" ]	TokenMizer: Graph-Structured Session Memory for Long-Horizon LLM Context Management	Large language model (LLM) deployments for long-horizon tasks face a fundamental constraint: context windows are finite while productive work sessions are not. When history exceeds the Maximum Effective Context Window (MECW), critical structured information - architectural decisions, task transitions, file histories - ...	[ "Shweta Mishra" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.06337	https://arxiv.org/pdf/2606.06337v1	2606.06337	null	0	0	true	https://github.com/Shweta-Mishra-ai/tokenmizer	null	0.65
3efa0fdc00a0e4150bb839ee3809a71d68049f847528dea13c97074daf4ec915	[ "arxiv", "semantic_scholar" ]	Discourse-Role Labels as Presentation-Time Variables for Context Use in Language Models	Context-augmented language model systems often wrap supplied content with labels such as Reference:, Evidence:, Instruction:, Note:, or Example:, but the effect of these labels on reader-model behavior remains underexplored. We introduce a paired fixed-content probe over 500 MMLU-Pro items: each item receives the same ...	[ "Jianguo Zhu", "Xiangmei Li", "Wenjie Liu" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.04109	https://arxiv.org/pdf/2606.04109v2	2606.04109	null	0	0	false	null	null	0.35
6eb6c898f43cade2472b10c0f271b1dde0a8672b452f9dcac14a90e600ec0519	[ "arxiv", "semantic_scholar" ]	Rethinking the Role of Positional Encoding: Sliding-Window Transformers without PE Remain Turing Complete	Positional encoding (PE) is widely viewed as necessary for transformers to process ordered sequences: without them, the next-token map appears permutation-invariant in its context tokens. This intuition underlies all prior universality results, which rely on positional information to prove that transformers with chain-...	[ "Qian Li", "Xinyu Mao", "Shang-Hua Teng" ]	[ "cs.LG", "cs.CC" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.01532	https://arxiv.org/pdf/2606.01532v2	2606.01532	null	0	0	false	null	null	0.35
51a418865f6983f13e05e889dc0522a3f96575ff62c0f54b4767b417a2a65337	[ "arxiv", "semantic_scholar" ]	Why Do Time Series Models Need Long Context Windows?	Modern deep learning models for forecasting groups of time series rely on increasingly longer observation windows. However, the benefit of increasing the window size is often simply attributed to capturing long-range dependencies, and broader discussion on how global forecasting models leverage input observations has b...	[ "Luca Butera", "Giovanni De Felice", "Andrea Cini", "Cesare Alippi" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.01999	https://arxiv.org/pdf/2606.01999v1	2606.01999	null	0	0	false	null	null	0.35
d5db30a2e4f53b7169428137f37e940e069c21a4932c72a471c7d9f074306915	[ "arxiv", "semantic_scholar" ]	LongAttnComp: Cross-Family Context Compression for Long-Context Reasoning	As real-world applications increasingly require processing inputs of 100k+ tokens, the gap between context length and inference efficiency has become a critical bottleneck. Context compression offers a way to reduce prefill costs while preserving task accuracy. However, existing training-free attention-based methods le...	[ "Mengmeng Ji", "Ravi Shanker Raju", "Jonathan Lingjie Li", "Chen Wu" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-31T00:00:00	https://arxiv.org/abs/2606.01336	https://arxiv.org/pdf/2606.01336v1	2606.01336	null	0	0	false	null	null	0.35
31e3870c5df401528dc775e238c269af506d00fde96d73c72a564ae634dedcc9	[ "arxiv", "semantic_scholar" ]	Periodic RoPE for Infinite Context LLMs	The ability to process ultra-long contexts is crucial for large language models (LLMs) to perform long-horizon tasks. While recent efforts have extended context windows to 1M and beyond, model performance degrades when sequence length exceeds the pre-trained range of positional encodings (e.g., RoPE), i.e., position ex...	[ "Simin Huo" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.27980	https://arxiv.org/pdf/2605.27980v1	2605.27980	null	0	0	true	https://github.com/Cominder/miniwin}{https://github.com/Cominder/miniwin}	null	0.65
c552cf680034fee798c7cabbde55a99f6aa22e969ce04a03069ddb68c7ddc2fa	[ "arxiv", "semantic_scholar" ]	BCER Agent: Reliable Long-Horizon MRI Workflow Execution via Compilation, Artifact Binding, and Bounded Local Recovery	Many recent medical VLM and agent studies are benchmarked on 2D images or comparatively short tool-calling exchanges, whereas real MRI analysis typically demands long, interdependent pipelines that operate on 3D/4D volumetric data. Under these conditions, reactive tool-calling agents are prone to cascading breakdowns t...	[ "Ziyang Long", "Xinqi Li", "Junzhou Chen", "Yifan Gao", "Debiao Li", "Hsin-Jung Yang" ]	[ "eess.IV" ]	[ "Engineering" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.29163	https://arxiv.org/pdf/2605.29163v1	2605.29163	null	0	0	true	https://github.com/Albertlongzi/BCER	null	0.65
383bb6e4f25169577320c864636bad41346fe4215e0e9e2b12fb2e35fc482fb8	[ "arxiv", "semantic_scholar" ]	Device Context Protocol: A Compact, Safety-First Architecture for LLM-Driven Control of Constrained Devices	Large language models are increasingly used as orchestrators of external tools via the Model Context Protocol (MCP), but MCP is built for software services with megabytes of memory and does not descend to the microcontrollers that dominate the long tail of physical devices. Recent work (IoT-MCP) ports MCP to edge gatew...	[ "Dongxu Yang" ]	[ "cs.NI", "cs.CR", "cs.LG" ]	[ "Computer Science" ]	2026-05-24T00:00:00	https://arxiv.org/abs/2605.26159	https://arxiv.org/pdf/2605.26159v1	2605.26159	null	0	0	true	https://github.com/device-context-protocol/dcp	null	0.65
93d694deb30347c0e02071d7b97e1948a4736d9baf3948cd96747eac67dca9e4	[ "arxiv", "semantic_scholar" ]	Positional Failures in Long-Context LLMs: A Blind Spot in Reasoning Benchmarks	Position-controlled evaluation is standard for retrieval tasks such as Needle-in-a-Haystack and RULER, but mainstream reasoning benchmarks do not control positional placement of target tasks in long contexts. We audit 11 long-context benchmarks and find none jointly controls task position, filler content, and context l...	[ "Chuyifei Zhang", "Hongyu Cui", "Xiaowen Huang", "Jitao Sang" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-05-22T00:00:00	https://arxiv.org/abs/2605.23170	https://arxiv.org/pdf/2605.23170v1	2605.23170	null	0	0	false	null	null	0.35
497a8671d278d8e204f3a77c46b5eb3d310e3bd7759f29ea6d225f24f1c3ae47	[ "arxiv", "semantic_scholar" ]	Parallel Context Compaction for Long-Horizon LLM Agent Serving	Long-horizon LLM agents accumulate growing conversation histories that eventually exceed the model's context window. Context compaction via LLM-based summarization keeps the conversation bounded, but summarization is inherently lossy and the blocking call stalls agent inference for tens of seconds. Moreover, the operat...	[ "Musa Cim", "Burak Topcu", "Chita Das", "Mahmut Taylan Kandemir" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-22T00:00:00	https://arxiv.org/abs/2605.23296	https://arxiv.org/pdf/2605.23296v1	2605.23296	null	0	0	false	null	null	0.35
c6f325268f898ab9ea22fa7c100adfb96c955a21b95e699db9d8f109a536d8fe	[ "arxiv", "semantic_scholar" ]	More Context, Larger Models, or Moral Knowledge? A Systematic Study of Schwartz Value Detection in Political Texts	Detecting Schwartz values in political text is difficult because implicit cues often depend on surrounding arguments and fine-grained distinctions between neighboring values. We study when context and explicit moral knowledge help sentence-level value detection. Using the ValuesML/Touché ValueEval format, we compare se...	[ "Víctor Yeste", "Paolo Rosso" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-05-21T00:00:00	https://arxiv.org/abs/2605.22641	https://arxiv.org/pdf/2605.22641v3	2605.22641	null	0	0	true	https://github.com/VictorMYeste/human-value-detection-context-rag	null	0.65
f4bf3018956b0344e469527a33116434e59b5b8640d9e521ada80b107afd24d9	[ "arxiv", "semantic_scholar" ]	The Efficiency Frontier: A Unified Framework for Cost-Performance Optimization in LLM Context Management	Large language models (LLMs) increasingly rely on long-context processing, but expanding context windows introduces substantial computational and financial costs. Existing context reduction approaches, including retrieval and memory compression methods, are typically evaluated using performance and efficiency metrics i...	[ "Binqi Shen", "Lier Jin", "Hanyu Cai", "Lan Hu", "Yuting Xin" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-21T00:00:00	https://arxiv.org/abs/2605.23071	https://arxiv.org/pdf/2605.23071v1	2605.23071	null	0	0	false	null	null	0.35
5816c4e746ef05b96975cf5b07890fab2f6065bd090728644507a8557ae7adad	[ "arxiv", "semantic_scholar" ]	PEEK: Context Map as an Orientation Cache for Long-Context LLM Agents	Large language model (LLM) agents increasingly operate over long and recurring external contexts, like document corpora and code repositories. Across invocations, existing approaches preserve either the agent's trajectory, passive access to raw material, or task-level strategies. None of them preserves what we argue is...	[ "Zhuohan Gu", "Qizheng Zhang", "Omar Khattab", "Samuel Madden" ]	[ "cs.AI", "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-05-19T00:00:00	https://arxiv.org/abs/2605.19932	https://arxiv.org/pdf/2605.19932v1	2605.19932	null	0	0	false	null	null	0.35
7782628a97eff9a8ec1939ee0a84633ed66e37baf1d2335a457b7ba482af4f3b	[ "arxiv", "semantic_scholar" ]	RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably	We identify intrinsic limitations of Rotary Positional Embeddings (RoPE) in Transformer-based long-context language models. Our theoretical analysis abstracts away from the specific content of the context and depends only on its length. We prove that as context length increases, RoPE-based attention becomes unpredictab...	[ "Yufeng Du", "Phillip Harris", "Minyang Tian", "Eliu A Huerta", "Srikanth Ronanki", "Subendhu Rongali", "Aram Galstyan", "Hao Peng" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-05-15T00:00:00	https://arxiv.org/abs/2605.15514	https://arxiv.org/pdf/2605.15514v1	2605.15514	null	0	0	false	null	null	0.35
ef126b35cc7f6a744dbf0633a98b0bb0545d77b6443e4bd0c9a551bf0ed20ede	[ "arxiv", "semantic_scholar" ]	EndPrompt: Efficient Long-Context Extension via Terminal Anchoring	Extending the context window of large language models typically requires training on sequences at the target length, incurring quadratic memory and computational costs that make long-context adaptation expensive and difficult to reproduce. We propose EndPrompt, a method that achieves effective context extension using o...	[ "Han Tian", "Luxuan Chen", "Xinran Chen", "Rui Kong", "Fang Wang", "Jiamin Chen", "Jinman Zhao", "Yuchen Li", "Jiashu Zhao", "Shuaiqiang Wang", "Haoyi Xiong", "Linghe Kong", "Dawei Yin" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-14T00:00:00	https://arxiv.org/abs/2605.14589	https://arxiv.org/pdf/2605.14589v2	2605.14589	null	0	0	true	https://github.com/clx1415926/EndPrompt	null	0.65
b86ebb5c5d283a527243e6a91370e150043334d8dd92ccda36d0bf567d8c55be	[ "arxiv", "semantic_scholar" ]	MetaBackdoor: Exploiting Positional Encoding as a Backdoor Attack Surface in LLMs	Backdoor attacks pose a serious security threat to large language models (LLMs), which are increasingly deployed as general-purpose assistants in safety- and privacy-critical applications. Existing LLM backdoors rely primarily on content-based triggers, requiring explicit modification of the input text. In this work, w...	[ "Rui Wen", "Mark Russinovich", "Andrew Paverd", "Jun Sakuma", "Ahmed Salem" ]	[ "cs.CR", "cs.CL" ]	[ "Computer Science" ]	2026-05-14T00:00:00	https://arxiv.org/abs/2605.15172	https://arxiv.org/pdf/2605.15172v1	2605.15172	null	0	0	false	null	null	0.35
f6ac71bf4effc1c63d60fc7d6cebf4345752e7d28243bdab928215bb8ce40d07	[ "arxiv", "semantic_scholar" ]	Correctness-Aware Repository Filtering Under Maximum Effective Context Window Constraints	Context window efficiency is a practical constraint in large language model (LLM)-based developer tools. Paulsen [12] shows that all tested models degrade in accuracy well before their advertised context limits the Maximum Effective Context Window (MECW) which makes context construction a quality problem, not just a co...	[ "Shweta Mishra" ]	[ "cs.SE", "cs.AI" ]	[ "Computer Science" ]	2026-05-14T00:00:00	https://arxiv.org/abs/2605.14362	https://arxiv.org/pdf/2605.14362v1	2605.14362	null	0	0	true	null	null	0.65
0874101642345ba3c1df1f096bdd66e9ec846e1e2c2c8529563fa69185497108	[ "arxiv", "semantic_scholar" ]	Training Long-Context Vision-Language Models Effectively with Generalization Beyond 128K Context	Long-context modeling is becoming a core capability of modern large vision-language models (LVLMs), enabling sustained context management across long-document understanding, video analysis, and multi-turn tool use in agentic workflows. Yet practical training recipes remain insufficiently explored, particularly for desi...	[ "Zhaowei Wang", "Lishu Luo", "Haodong Duan", "Weiwei Liu", "Sijin Wu", "Ji Luo", "Shen Yan", "Shuai Peng", "Sihang Yuan", "Chaoyi Huang", "Yi Lin", "Yangqiu Song" ]	[ "cs.CV" ]	[ "Computer Science" ]	2026-05-13T00:00:00	https://arxiv.org/abs/2605.13831	https://arxiv.org/pdf/2605.13831v1	2605.13831	null	0	0	false	null	null	0.35
b791e7395a39d5fbe8166baccd4a8b321a96cbae824eace0a34ccbdb03f08d54	[ "arxiv", "semantic_scholar" ]	Where Does Long-Context Supervision Actually Go? Effective-Context Exposure Balancing	Long-context adaptation is often viewed as window scaling, but this misses a token-level supervision mismatch: in packed training with document masking, each target token's effective context remains short. We introduce EXACT, a supervision-allocation objective that assigns extra weight to long effective-context targets...	[ "Jinchang Zhu", "Jindong Li", "Chengyu Zou", "Rong Fu", "Chao Wang", "Haowei He", "Menglin Yang" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-11T00:00:00	https://arxiv.org/abs/2605.10544	https://arxiv.org/pdf/2605.10544v1	2605.10544	null	0	0	false	null	null	0.35
22da49e90d9bff6dd3793d4da3ed5bbe0dee0d8d7fd3184aa21f104a0dca4a2f	[ "arxiv", "semantic_scholar" ]	Drift is a Sampling Error: SNR-Aware Power Distributions for Long-Horizon Robotic Planning	Despite rapid progress in Vision-Language-Action (VLA) models for robotic control, instruction drift remains a persistent failure mode in long-horizon tasks. This paper reconceptualizes this phenomenon, positing that instruction drift is fundamentally a systematic sampling error: local greedy sampling is prone to colla...	[ "Kewei Chen", "Yayu Long", "Mingsheng Shang" ]	[ "cs.RO" ]	[ "Computer Science" ]	2026-05-10T00:00:00	https://arxiv.org/abs/2605.09537	https://arxiv.org/pdf/2605.09537v1	2605.09537	null	0	0	false	null	null	0.35
75f53c631b1f93b4fe9dec2699b9f5ad2deb4a720ca7e0165abead8142d3af45	[ "arxiv", "semantic_scholar" ]	HexiSeq: Accommodating Long Context Training of LLMs over Heterogeneous Hardware	Long-context training of large language models (LLMs) is commonly distributed with Context Parallelism (CP) and Head Parallelism (HP), but existing training systems largely assume homogeneous GPU meshes. This paper extends CP and HP to heterogeneous GPU clusters with mixed GPU models and non-uniform network bandwidths,...	[ "Yan Liang", "Youhe Jiang", "Ran Yan", "Binhang Yuan", "Wei Wang", "Chuan Wu" ]	[ "cs.DC" ]	[ "Computer Science" ]	2026-05-08T00:00:00	https://arxiv.org/abs/2605.07569	https://arxiv.org/pdf/2605.07569v1	2605.07569	null	0	0	false	null	null	0.35
ae2002bd2c7f42a54fa39bb1c4332334ef111bb2ef4d2488dac3de3a24c4f533	[ "arxiv", "semantic_scholar" ]	A$^2$RD: Agentic Autoregressive Diffusion for Long Video Consistency	Synthesizing consistent and coherent long video remains a fundamental challenge. Existing methods suffer from semantic drift and narrative collapse over long horizons. We present A$^2$RD, an Agentic Auto-Regressive Diffusion architecture that decouples creative synthesis from consistency enforcement. A$^2$RD formulates...	[ "Do Xuan Long", "Yale Song", "Min-Yen Kan", "Tomas Pfister", "Long T. Le" ]	[ "cs.CV", "cs.AI" ]	[ "Computer Science" ]	2026-05-07T00:00:00	https://arxiv.org/abs/2605.06924	https://arxiv.org/pdf/2605.06924v1	2605.06924	null	1	0	false	null	null	0.35
2eb4c473a924f72b2870b3edb5140d28111b1a47d0311ee56c4b2e6c0c1f80e3	[ "arxiv", "semantic_scholar" ]	A Language for Describing Agentic LLM Contexts	Large language models are increasingly used within larger systems ("LLM agents"). These make a sequence of LLM calls, each call providing the LLM with a combination of instructions, observations, and interaction history. The design of the encoded information and its structure play a central role in the quality of the r...	[ "Noga Peleg Pelc", "Gal A. Kaminka", "Yoav Goldberg" ]	[ "cs.AI", "cs.CL", "cs.MA", "cs.SE" ]	[ "Computer Science" ]	2026-05-03T00:00:00	https://arxiv.org/abs/2605.01920	https://arxiv.org/pdf/2605.01920v1	2605.01920	10.1145/3786335.3813126	0	0	false	null	CAIS '26: ACM Conference on AI and Agentic Systems, May 2026, San Jose, CA, USA	0.55
f260577b473d5f16dc842cffd962a69558bc9e9013030b84446d01cce6c8b0b9	[ "arxiv", "semantic_scholar" ]	Accelerating Long-Tail Generation in Synchronous RLHF Training via Adaptive Tensor Parallelism	Reinforcement Learning from Human Feedback (RLHF) has become a key post-training paradigm for improving model quality. However, the synchronous three-stage RLHF pipeline is often bottlenecked by the generation stage, where response-length skew causes the effective batch size to shrink rapidly during decoding, leaving G...	[ "Long Zhao", "Qinghe Wang", "Jiaan Zhu", "Youhui Bai", "Zewen Jin", "Chaoyi Ruan", "Shengnan Wang", "Cheng Li" ]	[ "cs.AI", "cs.DC" ]	[ "Computer Science" ]	2026-05-03T00:00:00	https://arxiv.org/abs/2605.23945	https://arxiv.org/pdf/2605.23945v1	2605.23945	null	0	0	false	null	null	0.35
8710db2ab7139228be5e2f7f6c9da5408d83c2883c7f9d698c8bb7e9d8f4ad81	[ "arxiv", "semantic_scholar" ]	Beyond Compaction: Structured Context Eviction for Long-Horizon Agents	We present Context Window Lifecycle (CWL), a context-management scheme that gives long-horizon LLM agents an effectively unbounded working horizon. As a session accumulates history, CWL keeps the context within budget through graduated, semantically-aware eviction: the agent annotates its trajectory as typed, dependenc...	[ "Andrew Semenov", "Svyatoslav Dorofeev" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-01T00:00:00	https://arxiv.org/abs/2606.11213	https://arxiv.org/pdf/2606.11213v1	2606.11213	null	0	0	false	null	null	0.35
f752c981e458685d634226e67b38880aa4cc6c9e2d40a0e471bd1808a658c423	[ "arxiv", "semantic_scholar" ]	AutoSP: Unlocking Long-Context LLM Training Via Compiler-Based Sequence Parallelism	Large-language-models (LLMs) demonstrate enormous utility in long-context tasks which require processing prompts that consist of tens to hundreds of thousands of tokens. However, existing LLM training libraries do not provide easy to use abstractions to optimize for long-context training, instead focusing on optimizati...	[ "Ahan Gupta", "Zhihao Wang", "Neel Dani", "Masahiro Tanaka", "Olatunji Ruwase", "Minjia Zhang" ]	[ "cs.LG", "cs.DC", "cs.PF" ]	[ "Computer Science" ]	2026-04-29T00:00:00	https://arxiv.org/abs/2604.27089	https://arxiv.org/pdf/2604.27089v1	2604.27089	10.48550/arXiv.2604.27089	1	0	false	null	arXiv.org	0.55
49f072ed4ad7c6e2f1cff23269ee166e31ca501cdc85b10602bbc87a1dc4d823	[ "arxiv", "semantic_scholar" ]	Long-Context Aware Upcycling: A New Frontier for Hybrid LLM Scaling	Hybrid sequence models that combine efficient Transformer components with linear sequence modeling blocks are a promising alternative to pure Transformers, but most are still pretrained from scratch and therefore fail to reuse existing Transformer checkpoints. We study upcycling as a practical path to convert pretraine...	[ "Parsa Ashrafi Fashi", "Utkarsh Saxena", "Mehdi Rezagholizadeh", "Aref Jafari", "Akash Haridas", "Mingyu Yang", "Vansh Bhatia", "Guihong Li", "Vikram Appia", "Emad Barsoum" ]	[ "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-04-27T00:00:00	https://arxiv.org/abs/2604.24715	https://arxiv.org/pdf/2604.24715v1	2604.24715	10.48550/arXiv.2604.24715	0	0	false	null	arXiv.org	0.55
c5de05ed9f2f57d90fe0dfb60123c3371fc59b76550df56b647a09effe58e285	[ "arxiv", "semantic_scholar" ]	When Context Sticks: Studying Interference in In-Context Learning	This paper investigates context stickiness in in-context learning (ICL), a phenomenon where earlier examples in a prompt interfere with a transformer's ability to adapt to later tasks. Using synthetic regression tasks over linear and quadratic functions, we examine how models trained under sequential, mixed, and random...	[ "Hanna Rød", "Dagny Streit", "Nils Valseth Selte", "Justin Li" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-04-25T00:00:00	https://arxiv.org/abs/2604.23371	https://arxiv.org/pdf/2604.23371v1	2604.23371	10.48550/arXiv.2604.23371	0	0	true	https://github.com/nilsvselte/icl-context-stickiness	arXiv.org	0.85
c8463eb25f28d9f99c918065202aa32a3432ce44bb8adc1602cd7e304fb931b8	[ "arxiv", "semantic_scholar" ]	Shuffle the Context: RoPE-Perturbed Self-Distillation for Long-Context Adaptation	Large language models (LLMs) increasingly operate in settings that require reliable long-context understanding, such as retrieval-augmented generation and multi-document reasoning. A common strategy is to fine-tune pretrained short-context models at the target sequence length. However, we find that standard long-contex...	[ "Zichong Li", "Chen Liang", "Liliang Ren", "Tuo Zhao", "Yelong Shen", "Weizhu Chen" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-04-15T00:00:00	https://arxiv.org/abs/2604.14339	https://arxiv.org/pdf/2604.14339v1	2604.14339	10.48550/arXiv.2604.14339	0	0	false	null	arXiv.org	0.5443
f9b6961c4ef97aa9000b95c9ad12a43d52b53dc54244eb9ea3ba53c3d6abc383	[ "arxiv", "semantic_scholar" ]	Context Kubernetes: Declarative Orchestration of Enterprise Knowledge for Agentic AI Systems	We introduce Context Kubernetes, an architecture for orchestrating enterprise knowledge in agentic AI systems, with a prototype implementation and eight experiments. The core observation is that delivering the right knowledge, to the right agent, with the right permissions, at the right freshness -- across an entire or...	[ "Charafeddine Mouzouni" ]	[ "cs.AI", "cs.SE" ]	[ "Computer Science" ]	2026-04-13T00:00:00	https://arxiv.org/abs/2604.11623	https://arxiv.org/pdf/2604.11623v3	2604.11623	10.48550/arXiv.2604.11623	0	0	true	https://github.com/Cohorte-ai/context-kubernetes	arXiv.org	0.8376
f91f236995a09546fde12f5f6ee1ba0272e69b9aeff68171170891668045e568	[ "arxiv", "semantic_scholar" ]	A Decomposition Perspective to Long-context Reasoning for LLMs	Long-context reasoning is essential for complex real-world applications, yet remains a significant challenge for Large Language Models (LLMs). Despite the rapid evolution in long-context reasoning, current research often overlooks the internal complexity of the long-context reasoning task itself. In this paper, we move...	[ "Yanling Xiao", "Huaibing Xie", "Guoliang Zhao", "Shihan Dou", "Shaolei Wang", "Yiting Liu", "Nantao Zheng", "Cheng Zhang", "Pluto Zhou", "Zhisong Zhang", "Lemao Liu" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-04-09T00:00:00	https://arxiv.org/abs/2604.07981	https://arxiv.org/pdf/2604.07981v1	2604.07981	10.48550/arXiv.2604.07981	0	0	false	null	arXiv.org	0.5374
31071bafec02176b29d2a241d2521acfcf80f81fe9206c562857fbae5a07af21	[ "arxiv", "semantic_scholar" ]	Video-guided Machine Translation with Global Video Context	Video-guided Multimodal Translation (VMT) has advanced significantly in recent years. However, most existing methods rely on locally aligned video segments paired one-to-one with subtitles, limiting their ability to capture global narrative context across multiple segments in long videos. To overcome this limitation, w...	[ "Jian Chen", "JinZe Lv", "Zi Long", "XiangHua Fu" ]	[ "cs.CV", "cs.CL" ]	[ "Computer Science" ]	2026-04-08T00:00:00	https://arxiv.org/abs/2604.06789	https://arxiv.org/pdf/2604.06789v1	2604.06789	10.48550/arXiv.2604.06789	0	0	false	null	arXiv.org	0.5363
45da9b22671cb639e1c1a163e3ce08ec92c3948c9c9a89a668bf4ed448dc80ab	[ "arxiv", "semantic_scholar" ]	Short Data, Long Context: Distilling Positional Knowledge in Transformers	Extending the context window of language models typically requires expensive long-context pre-training, posing significant challenges for both training efficiency and data collection. In this paper, we present evidence that long-context retrieval capabilities can be transferred to student models through logit-based kno...	[ "Patrick Huber", "Ernie Chang", "Chinnadhurai Sankar", "Rylan Conway", "Igor Fedorov", "Md Rifat Arefin", "Adithya Sagar" ]	[ "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-04-07T00:00:00	https://arxiv.org/abs/2604.06070	https://arxiv.org/pdf/2604.06070v1	2604.06070	10.48550/arXiv.2604.06070	0	0	false	null	arXiv.org	0.5351
4666ed60456a709d99e6cad627d1cf252e251716502708f0d64c9ddd2dbcdc30	[ "arxiv", "semantic_scholar" ]	$π^2$: Structure-Originated Reasoning Data Improves Long-Context Reasoning Ability of Large Language Models	We study a pipeline that curates reasoning data from initial structured data for improving long-context reasoning in large language models (LLMs). Our approach, $π^2$, constructs high-quality reasoning data through rigorous QA curation: 1) extracting and expanding tables from Wikipedia, 2) from the collected tables and...	[ "Quyet V. Do", "Thinh Pham", "Nguyen Nguyen", "Sha Li", "Pratibha Zunjare", "Tu Vu" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-04-06T00:00:00	https://arxiv.org/abs/2604.05114	https://arxiv.org/pdf/2604.05114v1	2604.05114	10.48550/arXiv.2604.05114	1	0	true	https://github.com/vt-pi-squared/pi-squared	arXiv.org	0.8252
9962d7de9b2e43fa10f41b4d1fe11e55d64f1408698e58e54c18967d8871e3f2	[ "arxiv", "semantic_scholar" ]	Long-Context Reasoning Through Proxy-Based Chain-of-Thought Tuning	Recent large language models support inputs of up to 10 million tokens, yet they perform poorly on long-context tasks that require complex reasoning. Such tasks can be solved using only a subset of the input -- a proxy context -- rather than the full sequence. Despite sharing the same underlying reasoning process, mode...	[ "Miao Li", "Irina Saparina", "Alexander Gurung", "Mirella Lapata" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-04-06T00:00:00	https://arxiv.org/abs/2605.20201	https://arxiv.org/pdf/2605.20201v2	2605.20201	null	0	0	false	null	null	0.3398
8842c6f5c655103f16dda518954c1860b810b8b58affdbd3d2a7e35e37ed11d6	[ "arxiv", "semantic_scholar" ]	Long-Reach Robotic Manipulation for Assembly and Outfitting of Lunar Structures	Future infrastructure construction on the lunar surface will require semi- or fully-autonomous operation from robots deployed at the build site. In particular, tasks such as electrical outfitting necessitate transport, routing, and fine manipulation of cables across large structures. To address this need, we present a ...	[ "Stanley Wang", "Venny Kojouharov", "Long Yin Chung", "Daniel Morton", "Mark Cutkosky" ]	[ "cs.RO" ]	[ "Computer Science" ]	2026-03-31T00:00:00	https://arxiv.org/abs/2603.29226	https://arxiv.org/pdf/2603.29226v1	2603.29226	10.1109/iSpaRo66239.2025.11437024	2	0	false	null	null	0.3354
31616d6178f023c384fd8fbbd40e3c871003dcee94fd27a7f2ac55f8dc79026d	[ "arxiv", "semantic_scholar" ]	Long-Reach Robotic Cleaning for Lunar Solar Arrays	Commercial lunar activity is accelerating the need for reliable surface infrastructure and routine operations to keep it functioning. Maintenance tasks such as inspection, cleaning, dust mitigation, and minor repair are essential to preserve performance and extend system life. A specific application is the cleaning of ...	[ "Stanley Wang", "Velin Kojouharov", "Long Yin Chung", "Daniel Morton", "Mark Cutkosky" ]	[ "cs.RO" ]	[ "Computer Science" ]	2026-03-31T00:00:00	https://arxiv.org/abs/2603.29240	https://arxiv.org/pdf/2603.29240v1	2603.29240	10.48550/arXiv.2603.29240	0	0	false	null	arXiv.org	0.5271
8997a92a1e89e115cc6b4941c5f8a918fad3ce3ffe0381820e4f4d7027fd7fdc	[ "arxiv", "semantic_scholar" ]	Developing Adaptive Context Compression Techniques for Large Language Models (LLMs) in Long-Running Interactions	Large Language Models (LLMs) often experience performance degradation during long-running interactions due to increasing context length, memory saturation, and computational overhead. This paper presents an adaptive context compression framework that integrates importance-aware memory selection, coherence-sensitive fil...	[ "Payal Fofadiya", "Sunil Tiwari" ]	[ "cs.CV", "cs.AI" ]	[ "Computer Science" ]	2026-03-31T00:00:00	https://arxiv.org/abs/2603.29193	https://arxiv.org/pdf/2603.29193v1	2603.29193	10.48550/arXiv.2603.29193	0	0	false	null	arXiv.org	0.5271
dd6c0db625bbf56eff3a3d846ec56e4d3177ecd019b45e063828c1f3d6b14568	[ "arxiv", "semantic_scholar" ]	MemoryCD: Benchmarking Long-Context User Memory of LLM Agents for Lifelong Cross-Domain Personalization	Recent advancements in Large Language Models (LLMs) have expanded context windows to million-token scales, yet benchmarks for evaluating memory remain limited to short-session synthetic dialogues. We introduce \textsc{MemoryCD}, the first large-scale, user-centric, cross-domain memory benchmark derived from lifelong re...	[ "Weizhi Zhang", "Xiaokai Wei", "Wei-Chieh Huang", "Zheng Hui", "Chen Wang", "Michelle Gong", "Philip S. Yu" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-03-26T00:00:00	https://arxiv.org/abs/2603.25973	https://arxiv.org/pdf/2603.25973v1	2603.25973	10.48550/arXiv.2603.25973	7	0	false	null	arXiv.org	0.5214
cde7954fa97265f240e722b727f09ea65fcb6a74fe626a36b5ee89ffa55ef0c7	[ "arxiv", "semantic_scholar" ]	Reasoner-Executor-Synthesizer: Scalable Agentic Architecture with Static O(1) Context Window	Large Language Models (LLMs) deployed as autonomous agents commonly use Retrieval-Augmented Generation (RAG), feeding retrieved documents into the context window, which creates two problems: the risk of hallucination grows with context length, and token cost scales linearly with dataset size. We propose the Reasoner-Ex...	[ "Ivan Dobrovolskyi" ]	[ "cs.IR", "cs.AI" ]	[ "Computer Science" ]	2026-03-23T00:00:00	https://arxiv.org/abs/2603.22367	https://arxiv.org/pdf/2603.22367v1	2603.22367	10.48550/arXiv.2603.22367	0	0	false	null	arXiv.org	0.5179
99df631ca195ec785bd83ac377fb879f8a08eb0a8e4615a190ef5ba8244525bd	[ "arxiv", "semantic_scholar" ]	Conversation Tree Architecture: A Structured Framework for Context-Aware Multi-Branch LLM Conversations	Large language models (LLMs) are increasingly deployed for extended, multi-topic conversations, yet the flat, append-only structure of current conversation interfaces introduces a fundamental limitation: all context accumulates in a single unbounded window, causing topically distinct threads to bleed into one another a...	[ "Pranav Hemanth", "Sampriti Saha" ]	[ "cs.CL", "cs.AI", "cs.HC" ]	[ "Computer Science" ]	2026-03-22T00:00:00	https://arxiv.org/abs/2603.21278	https://arxiv.org/pdf/2603.21278v1	2603.21278	10.48550/arXiv.2603.21278	0	0	false	null	arXiv.org	0.5168
6718abf84cfe2befedec977483c4fd97115bef34ca52f24f73ad93a65967527d	[ "arxiv", "semantic_scholar" ]	MKA: Memory-Keyed Attention for Efficient Long-Context Reasoning	As long-context language modeling becomes increasingly important, the cost of maintaining and attending to large Key/Value (KV) caches grows rapidly, becoming a major bottleneck in both training and inference. While prior works such as Multi-Query Attention (MQA) and Multi-Latent Attention (MLA) reduce memory by sharin...	[ "Dong Liu", "Yanxuan Yu", "Ben Lengerich", "Ying Nian Wu" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-03-21T00:00:00	https://arxiv.org/abs/2603.20586	https://arxiv.org/pdf/2603.20586v2	2603.20586	10.48550/arXiv.2603.20586	9	0	false	null	arXiv.org	0.5156
511989ccf4603e4547100d47c61e64e9297ec8a841b3226201e351751239b4fe	[ "arxiv", "semantic_scholar" ]	Coding Agents are Effective Long-Context Processors	Large Language Models (LLMs) have demonstrated remarkable progress in scaling to access massive contexts. However, the access is via the latent and uninterpretable attention mechanisms, and LLMs fail to effective process long context, exhibiting significant performance degradation as context length increases. In this w...	[ "Weili Cao", "Xunjian Yin", "Bhuwan Dhingra", "Shuyan Zhou" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-03-20T00:00:00	https://arxiv.org/abs/2603.20432	https://arxiv.org/pdf/2603.20432v1	2603.20432	10.48550/arXiv.2603.20432	5	1	false	null	arXiv.org	0.5145
be1dd14fa6bcf09ff067044566d4b9b42556b3b8c0ab9121fa4db1d102889246	[ "arxiv", "semantic_scholar" ]	The $\mathbf{Y}$-Combinator for LLMs: Solving Long-Context Rot with $λ$-Calculus	LLMs are increasingly used as general-purpose reasoners, but long inputs remain bottlenecked by a fixed context window. Recursive Language Models (RLMs) address this by externalising the prompt and recursively solving subproblems. Yet existing RLMs depend on an open-ended read-eval-print loop (REPL) in which the model ...	[ "Amartya Roy", "Rasul Tutunov", "Xiaotong Ji", "Matthieu Zimmer", "Haitham Bou-Ammar" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-03-20T00:00:00	https://arxiv.org/abs/2603.20105	https://arxiv.org/pdf/2603.20105v1	2603.20105	10.48550/arXiv.2603.20105	1	0	true	https://github.com/lambda-calculus-LLM/lambda-RLM	arXiv.org	0.7951
e1314b19696e108579b7e427d2e0c936f4958d017fa36c278959c440bc036d2e	[ "arxiv", "semantic_scholar" ]	UT-ACA: Uncertainty-Triggered Adaptive Context Allocation for Long-Context Inference	Long-context inference remains challenging for large language models due to attention dilution and out-of-distribution degradation. Context selection mitigates this limitation by attending to a subset of key-value cache entries, yet most methods allocate a fixed context budget throughout decoding despite highly non-uni...	[ "Lang Zhou", "Shuxuan Li", "Zhuohao Li", "Shi Liu", "Zhilin Zhao", "Wei-Shi Zheng" ]	[ "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-03-19T00:00:00	https://arxiv.org/abs/2603.18446	https://arxiv.org/pdf/2603.18446v1	2603.18446	10.48550/arXiv.2603.18446	0	0	false	null	arXiv.org	0.5133
0009cffae66956b8054a21dea2bb6182cf480edb70affb48c3238ece66368293	[ "arxiv", "semantic_scholar" ]	Difference-Based High-Dimensional Long-Run Covariance Matrix Estimation for Mean-shift Time Series	We consider estimation of high-dimensional long-run covariance matrices for time series with nonconstant means, a setting in which conventional estimators can be severely biased. To address this difficulty, we propose a difference-based initial estimator that is robust to a broad class of mean variations, and combine i...	[ "Yanhong Liu", "Fengyi Song", "Long Feng" ]	[ "stat.ME" ]	[ "Mathematics" ]	2026-03-18T00:00:00	https://arxiv.org/abs/2603.17226	https://arxiv.org/pdf/2603.17226v1	2603.17226	null	0	0	false	null	null	0.3259
6a14dc39117b75261c88b20424b37b4fb359ad2195e959bc43fe2fae9c95be97	[ "arxiv", "semantic_scholar" ]	Causal Cellular Context Transfer Learning (C3TL): An Efficient Architecture for Prediction of Unseen Perturbation Effects	Predicting the effects of chemical and genetic perturbations on quantitative cell states is a central challenge in computational biology, molecular medicine and drug discovery. Recent work has leveraged large-scale single-cell data and massive foundation models to address this task. However, such computational resource...	[ "Michael Scholkemper", "Sach Mukherjee" ]	[ "cs.LG", "q-bio.QM" ]	[ "Computer Science", "Biology" ]	2026-03-13T00:00:00	https://arxiv.org/abs/2603.13051	https://arxiv.org/pdf/2603.13051v1	2603.13051	10.48550/arXiv.2603.13051	0	0	false	null	arXiv.org	0.5065
f5a947fd476a2fd8378bb5d53f7226c4f43226d92a0a76c9279ca73faae6f313	[ "arxiv", "semantic_scholar" ]	StatePlane: A Cognitive State Plane for Long-Horizon AI Systems Under Bounded Context	Large language models (LLMs) and small language models (SLMs) operate under strict context window and key-value (KV) cache constraints, fundamentally limiting their ability to reason coherently over long interaction horizons. Existing approaches -- extended context windows, retrieval-augmented generation, summarization...	[ "Sasank Annapureddy", "John Mulcahy", "Anjaneya Prasad Thamatani" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-03-13T00:00:00	https://arxiv.org/abs/2603.13644	https://arxiv.org/pdf/2603.13644v1	2603.13644	10.48550/arXiv.2603.13644	1	0	false	null	arXiv.org	0.5065
16f31e7cd9b7bc7d45d796131b859aed010b69c5e39866d31637c43c364dc2be	[ "arxiv", "semantic_scholar" ]	Long-Context Encoder Models for Polish Language Understanding	While decoder-only Large Language Models (LLMs) have recently dominated the NLP landscape, encoder-only architectures remain a cost-effective and parameter-efficient standard for discriminative tasks. However, classic encoders like BERT are limited by a short context window, which is insufficient for processing long do...	[ "Sławomir Dadas", "Rafał Poświata", "Marek Kozłowski", "Małgorzata Grębowiec", "Michał Perełkiewicz", "Paweł Klimiuk", "Przemysław Boruta" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-03-12T00:00:00	https://arxiv.org/abs/2603.12191	https://arxiv.org/pdf/2603.12191v1	2603.12191	10.48550/arXiv.2603.12191	1	0	false	null	arXiv.org	0.5053
d5ce7518569ed8538541977e2ee11fe788fdeecdd4e4a5e1c98712d6de8ad1e8	[ "arxiv", "semantic_scholar" ]	The Missing Memory Hierarchy: Demand Paging for LLM Context Windows	The context window of a large language model is not memory. It is L1 cache: a small, fast, expensive resource that the field treats as the entire memory system. There is no L2, no virtual memory, no paging. Every tool definition, every system prompt, and every stale tool result occupies context for the lifetime of the ...	[ "Tony Mason" ]	[ "cs.OS", "cs.AI", "cs.SE" ]	[ "Computer Science" ]	2026-03-09T00:00:00	https://arxiv.org/abs/2603.09023	https://arxiv.org/pdf/2603.09023v1	2603.09023	10.48550/arXiv.2603.09023	1	0	false	null	arXiv.org	0.5019
2b8a21d95c6bee8b94f4ee556b18da0eea35a8008336b3c770963fe98342358a	[ "arxiv", "semantic_scholar" ]	Stacked from One: Multi-Scale Self-Injection for Context Window Extension	The limited context window of contemporary large language models (LLMs) remains a primary bottleneck for their broader application across diverse domains. Although continual pre-training on long-context data offers a straightforward solution, it incurs prohibitive data acquisition and computational costs. To address th...	[ "Wei Han", "Pan Zhou", "Soujanya Poria", "Shuicheng Yan" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-03-05T00:00:00	https://arxiv.org/abs/2603.04759	https://arxiv.org/pdf/2603.04759v2	2603.04759	10.48550/arXiv.2603.04759	0	0	false	null	arXiv.org	0.4973
c8e0d3dea9f23866623e571254506854d6667c3ce002b9724f97f79dcd4a56b7	[ "arxiv", "semantic_scholar" ]	Beyond the Context Window: A Cost-Performance Analysis of Fact-Based Memory vs. Long-Context LLMs for Persistent Agents	Persistent conversational AI systems face a choice between passing full conversation histories to a long-context large language model (LLM) and maintaining a dedicated memory system that extracts and retrieves structured facts. We compare a fact-based memory system built on the Mem0 framework against long-context LLM i...	[ "Natchanon Pollertlam", "Witchayut Kornsuwannawit" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-03-05T00:00:00	https://arxiv.org/abs/2603.04814	https://arxiv.org/pdf/2603.04814v1	2603.04814	10.48550/arXiv.2603.04814	4	1	false	null	arXiv.org	0.4973
51de898e81b52c5238dccc58b5833b1a38699553fc50f6ad7d40f2f9bb076db9	[ "arxiv", "semantic_scholar" ]	Engaging students with statistics through choice of real data context on homework	Statistics educators recommend teaching with real data with relevant contexts, but defining relevancy is challenging and varies by student. We investigated whether providing student choice of data context increases engagement through a quasi-experiment in two sections of an introductory probability and statistics cours...	[ "Catalina Medina", "Mine Dogucu" ]	[ "stat.OT" ]	[ "Mathematics" ]	2026-03-04T00:00:00	https://arxiv.org/abs/2603.04541	https://arxiv.org/pdf/2603.04541v1	2603.04541	null	0	0	true	https://github.com/CatalinaMedina/data-context-choice-manuscript	null	0.5864
75bddd0f5b9c27d3377c5c56060eed52afdd8611f5f51f95b43ae1f719c3e6a6	[ "arxiv", "semantic_scholar" ]	Cross-Family Speculative Prefill: Training-Free Long-Context Compression with Small Draft Models	Prompt length is a major bottleneck in agentic large language model (LLM) workloads, where repeated inference steps and multi-call loops incur substantial prefill cost. Recent work on speculative prefill demonstrates that attention-based token importance estimation can enable training-free prompt compression, but this ...	[ "Shubhangi Upasani", "Ravi Shanker Raju", "Bo Li", "Mengmeng Ji", "John Long", "Chen Wu", "Urmish Thakker", "Guangtao Wang" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-03-03T00:00:00	https://arxiv.org/abs/2603.02631	https://arxiv.org/pdf/2603.02631v3	2603.02631	10.48550/arXiv.2603.02631	1	0	false	null	arXiv.org	0.495
cda5d857d46364efc3cc7d88650c1060475241a753e44356eea16e5e24d4d82a	[ "arxiv", "semantic_scholar" ]	An Evaluation of Context Length Extrapolation in Long Code via Positional Embeddings and Efficient Attention	The rapid advancement of large language models (LLMs) has led to a significant increase in automated tools in the software engineering, capable of performing various code-related tasks such as code generation, completion, and translation. Despite these advancements, its effectiveness is constrained by fixed context len...	[ "Madhusudan Ghosh", "Rishabh Gupta" ]	[ "cs.SE", "cs.AI" ]	[ "Computer Science" ]	2026-02-25T00:00:00	https://arxiv.org/abs/2602.21800	https://arxiv.org/pdf/2602.21800v1	2602.21800	10.48550/arXiv.2602.21800	0	0	false	null	arXiv.org	0.4881
413272986dce8c473b1378d4176d8fb74837b5510154f840be758bd58256dab0	[ "arxiv", "semantic_scholar" ]	Efficient Scaling of LLM Training with Flexible Context Parallelism	Scaling long-context capabilities is crucial for Large Language Models (LLMs). However, real-world data contain a large number of sequences with heterogeneous lengths. Existing training libraries for LLMs rely on static parallelism strategies, which suffer from severe load imbalance, redundant communication, and subopt...	[ "Yifan Niu", "Han Xiao", "Dongyi Liu", "Wei Zhou", "Jia Li" ]	[ "cs.DC", "cs.LG" ]	[ "Computer Science" ]	2026-02-25T00:00:00	https://arxiv.org/abs/2602.21788	https://arxiv.org/pdf/2602.21788v2	2602.21788	null	1	0	false	null	null	0.3106
c401b699dfe6ac4dce45d02635aa05f556db1d6c9abb94963c5f540e1f306965	[ "arxiv", "semantic_scholar" ]	Codified Context: Infrastructure for AI Agents in a Complex Codebase	LLM-based agentic coding assistants lack persistent memory: they lose coherence across sessions, forget project conventions, and repeat known mistakes. Recent studies characterize how developers configure agents through manifest files, but an open challenge remains how to scale such configurations for large, multi-agen...	[ "Aristidis Vasilopoulos" ]	[ "cs.SE" ]	[ "Computer Science" ]	2026-02-24T00:00:00	https://arxiv.org/abs/2602.20478	https://arxiv.org/pdf/2602.20478v1	2602.20478	10.48550/arXiv.2602.20478	5	1	true	https://github.com/arisvas4/codified-context-infrastructure	arXiv.org	0.7526
84f59786c029fe44b0337c996559b7747914ef397e3cdeee8345f350e4f10618	[ "arxiv", "semantic_scholar" ]	CHESS: Context-aware Hierarchical Efficient Semantic Selection for Long-Context LLM Inference	Long-context LLMs demand accurate inference at low latency, yet decoding becomes primarily constrained by KV cache as context grows. Prior pruning methods are largely context-agnostic: their token selection ignores step-wise relevance and local semantics, which undermines quality. Moreover, their irregular accesses and...	[ "Chao Fei", "Guozhong Li", "Chenxi Liu", "Panos Kalnis" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-02-24T00:00:00	https://arxiv.org/abs/2602.20732	https://arxiv.org/pdf/2602.20732v1	2602.20732	10.48550/arXiv.2602.20732	0	0	true	null	arXiv.org	0.7526
7eaae19db333ce5795413102d01660dbc4af6420cef1d14a556d4bc891e4cb60	[ "arxiv", "semantic_scholar" ]	FAST-Prefill: FPGA Accelerated Sparse Attention for Long Context LLM Prefill	In long-context large language model (LLM) inference, the prefill stage dominates computation due to self-attention over the complete input context. Sparse attention significantly reduces self-attention computation by limiting each token's interactions to a subset of tokens. The attention sparsity pattern varies across...	[ "Rakshith Jayanth", "Viktor Prasanna" ]	[ "cs.AR" ]	[ "Computer Science" ]	2026-02-24T00:00:00	https://arxiv.org/abs/2602.20515	https://arxiv.org/pdf/2602.20515v1	2602.20515	10.1109/FCCM68464.2026.00067	0	0	false	null	IEEE Symposium on Field-Programmable Custom Computing Machines	0.487
83bb72a6c671955abcd9c00198968afa356cba61c217aa763d83d05792ce160e	[ "arxiv", "semantic_scholar" ]	The Convergence of Schema-Guided Dialogue Systems and the Model Context Protocol	This paper establishes a fundamental convergence: Schema-Guided Dialogue (SGD) and the Model Context Protocol (MCP) represent two manifestations of a unified paradigm for deterministic, auditable LLM-agent interaction. SGD, designed for dialogue-based API discovery (2019), and MCP, now the de facto standard for LLM-too...	[ "Andreas Schlapbach" ]	[ "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-02-21T00:00:00	https://arxiv.org/abs/2602.18764	https://arxiv.org/pdf/2602.18764v2	2602.18764	10.48550/arXiv.2602.18764	1	0	false	null	arXiv.org	0.4835
f5d78dd048be0bace346ad59c2439ae2d7373214385c3bd2fb4756925718e7c1	[ "arxiv", "semantic_scholar" ]	The Limits of Long-Context Reasoning in Automated Bug Fixing	Rapidly increasing context lengths have led to the assumption that large language models (LLMs) can directly reason over entire codebases. Concurrently, recent advances in LLMs have enabled strong performance on software engineering benchmarks, particularly when paired with agentic workflows. In this work, we systemati...	[ "Ravi Raju", "Mengmeng Ji", "Shubhangi Upasani", "Bo Li", "Urmish Thakker" ]	[ "cs.SE", "cs.LG" ]	[ "Computer Science" ]	2026-02-17T00:00:00	https://arxiv.org/abs/2602.16069	https://arxiv.org/pdf/2602.16069v2	2602.16069	10.48550/arXiv.2602.16069	1	0	true	null	arXiv.org	0.7402
fb4bc9491ead5aeec4e0cc8902c14df21ea19cca373121c598f087ff2a3e752c	[ "arxiv", "semantic_scholar" ]	Long Context, Less Focus: A Scaling Gap in LLMs Revealed through Privacy and Personalization	Large language models (LLMs) are increasingly deployed in privacy-critical and personalization-oriented scenarios, yet the role of context length in shaping privacy leakage and personalization effectiveness remains largely unexplored. We introduce a large-scale benchmark, PAPerBench, to systematically study how increas...	[ "Shangding Gu" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-02-16T00:00:00	https://arxiv.org/abs/2602.15028	https://arxiv.org/pdf/2602.15028v1	2602.15028	10.48550/arXiv.2602.15028	5	0	true	https://github.com/SafeRL-Lab/PAPerBench	arXiv.org	0.7384
ee5b8a1b03844136c9510a199b2b7f0eb5dda020710272f67c8904848ac7bdbf	[ "arxiv", "semantic_scholar" ]	GPT-5 vs Other LLMs in Long Short-Context Performance	With the significant expansion of the context window in Large Language Models (LLMs), these models are theoretically capable of processing millions of tokens in a single pass. However, research indicates a significant gap between this theoretical capacity and the practical ability of models to robustly utilize informat...	[ "Nima Esmi", "Maryam Nezhad-Moghaddam", "Fatemeh Borhani", "Asadollah Shahbahrami", "Amin Daemdoost", "Georgi Gaydadjiev" ]	[ "cs.CL", "cs.AI", "cs.HC" ]	[ "Computer Science" ]	2026-02-15T00:00:00	https://arxiv.org/abs/2602.14188	https://arxiv.org/pdf/2602.14188v1	2602.14188	10.1109/FLLM67465.2025.11391194	0	0	false	null	null	0.3033
14f13d3c68f93ce82c8008f5913dc1a0a762d433c5fe3bfd3d9d2c1d50133fe5	[ "arxiv", "semantic_scholar" ]	Rotary Positional Embeddings as Phase Modulation: Theoretical Bounds on the RoPE Base for Long-Context Transformers	Rotary positional embeddings (RoPE) are widely used in large language models to encode token positions through multiplicative rotations, yet their behavior at long context lengths remains poorly characterized. In this work, we reinterpret RoPE as phase modulation applied to a bank of complex oscillators, enabling analy...	[ "Feilong Liu" ]	[ "cs.LG", "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-02-11T00:00:00	https://arxiv.org/abs/2602.10959	https://arxiv.org/pdf/2602.10959v1	2602.10959	10.48550/arXiv.2602.10959	2	1	false	null	arXiv.org	0.4721
e1b6d5d4c993ed34494e78635a871f27d614c55c4cfd5cb6b15716b3e3ea34f3	[ "arxiv" ]	When Less is More: The LLM Scaling Paradox in Context Compression	Scaling up model parameters has long been a prevalent training paradigm driven by the assumption that larger models yield superior generation capabilities. However, under lossy context compression in a compressor--decoder setup, we find a \textbf{\textit{Size-Fidelity Paradox}}: increasing compressor size can lessen th...	[ "Ruishan Guo", "Yibing Liu", "Guoxin Ma", "Yan Wang", "Yueyang Zhang", "Long Xia", "Kecheng Chen", "Zhiyuan Sun", "Daiting Shi" ]	[ "cs.LG" ]	[]	2026-02-10T00:00:00	https://arxiv.org/abs/2602.09789	https://arxiv.org/pdf/2602.09789v3	2602.09789	null	0	0	false	null	null	0.2997
27e7fa1e4a41edec0b76c3ce0179d191cd76c6ad98095c9c9eb432271f187c42	[ "arxiv", "semantic_scholar" ]	Do Reasoning LLMs Refuse What They Infer in Long Contexts?	Long-context LLMs can infer objectives that are not stated explicitly. This capability is useful for reasoning over documents, code, retrieved evidence, and tool traces, but it also creates a safety risk: harmful intent can be distributed across a context and become visible only after the model composes the relevant pi...	[ "Yu Fu", "Haz Sameen Shahgir", "Huanli Gong", "Zhipeng Wei", "N. Benjamin Erichson", "Yue Dong" ]	[ "cs.CL", "cs.CR" ]	[ "Computer Science" ]	2026-02-09T00:00:00	https://arxiv.org/abs/2602.08874	https://arxiv.org/pdf/2602.08874v2	2602.08874	null	3	1	false	null	null	0.299
8ea0865f89f2e692b94f0e8602c5fa84d8369634ae8a9835b5f3844bdb6c15f8	[ "arxiv", "semantic_scholar" ]	Context Forcing: Consistent Autoregressive Video Generation with Long Context	Recent approaches to real-time long video generation typically employ streaming tuning strategies, attempting to train a long-context student using a short-context (memoryless) teacher. In these frameworks, the student performs long rollouts but receives supervision from a teacher limited to short 5-second windows. Thi...	[ "Shuo Chen", "Cong Wei", "Sun Sun", "Ping Nie", "Kai Zhou", "Ge Zhang", "Ming-Hsuan Yang", "Wenhu Chen" ]	[ "cs.CV" ]	[ "Computer Science" ]	2026-02-05T00:00:00	https://arxiv.org/abs/2602.06028	https://arxiv.org/pdf/2602.06028v1	2602.06028	10.48550/arXiv.2602.06028	24	1	false	null	arXiv.org	0.4652
0b5a70fb0622d4e0debeb8451906e5578f9cb5f2a6eef83a90433750838cd8af	[ "arxiv", "semantic_scholar" ]	Simulated Adoption: Decoupling Magnitude and Direction in LLM In-Context Conflict Resolution	Large Language Models (LLMs) frequently prioritize conflicting in-context information over pre-existing parametric memory, a phenomenon often termed sycophancy or compliance. However, the mechanistic realization of this behavior remains obscure, specifically how the model resolves these knowledge conflicts through comp...	[ "Long Zhang", "Fangwei Lin" ]	[ "cs.LG", "cs.CL", "cs.CY" ]	[ "Computer Science" ]	2026-02-04T00:00:00	https://arxiv.org/abs/2602.04918	https://arxiv.org/pdf/2602.04918v2	2602.04918	10.48550/arXiv.2602.04918	0	0	false	null	arXiv.org	0.4641
b0f2c8b875e126495561622b98c904e33b28b54775b4984e682f473f252452aa	[ "arxiv", "semantic_scholar" ]	LycheeDecode: Accelerating Long-Context LLM Inference via Hybrid-Head Sparse Decoding	The proliferation of long-context large language models (LLMs) exposes a key bottleneck: the rapidly expanding key-value cache during decoding, which imposes heavy memory and latency costs. While recent approaches attempt to alleviate this by sharing a single set of crucial tokens across layers, such coarse-grained sha...	[ "Gang Lin", "Dongfang Li", "Zhuoen Chen", "Yukun Shi", "Xuhui Chen", "Baotian Hu", "Min Zhang" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-02-04T00:00:00	https://arxiv.org/abs/2602.04541	https://arxiv.org/pdf/2602.04541v1	2602.04541	10.48550/arXiv.2602.04541	3	0	false	null	arXiv.org	0.4641
807269a479aa415a2eee04b41e12407345a60c7fcfa41bfd848d5b3269655a63	[ "arxiv", "semantic_scholar" ]	ATACompressor: Adaptive Task-Aware Compression for Efficient Long-Context Processing in LLMs	Long-context inputs in large language models (LLMs) often suffer from the "lost in the middle" problem, where critical information becomes diluted or ignored due to excessive length. Context compression methods aim to address this by reducing input size, but existing approaches struggle with balancing information prese...	[ "Xuancheng Li", "Haitao Li", "Yujia Zhou", "Qingyao Ai", "Yiqun Liu" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-02-03T00:00:00	https://arxiv.org/abs/2602.03226	https://arxiv.org/pdf/2602.03226v1	2602.03226	10.1145/3767695.3769499	5	1	false	null	null	0.2946
4cafd1d6d3c3c34db8799a0eff28e85946ea39a61f07d4838832a51ba32abf3d	[ "arxiv", "semantic_scholar" ]	Focus-dLLM: Accelerating Long-Context Diffusion LLM Inference via Confidence-Guided Context Focusing	Diffusion Large Language Models (dLLMs) deliver strong long-context processing capability in a non-autoregressive decoding paradigm. However, the considerable computational cost of bidirectional full attention limits the inference efficiency. Although sparse attention is promising, existing methods remain ineffective. ...	[ "Lingkun Long", "Yushi Huang", "Shihao Bai", "Ruihao Gong", "Jun Zhang", "Ao Zhou", "Jianlei Yang" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-02-02T00:00:00	https://arxiv.org/abs/2602.02159	https://arxiv.org/pdf/2602.02159v1	2602.02159	10.48550/arXiv.2602.02159	2	0	true	https://github.com/Longxmas/Focus-dLLM	arXiv.org	0.7136
46df33c3d74aa5f184d48530720b01ae1d1d36000a78bcde895c9c0c638059c3	[ "arxiv", "semantic_scholar" ]	Latent Context Compilation: Distilling Long Context into Compact Portable Memory	Efficient long-context LLM deployment is stalled by a dichotomy between amortized compression, which struggles with out-of-distribution generalization, and Test-Time Training, which incurs prohibitive synthetic data costs and requires modifying model weights, creating stateful parameters that complicate concurrent serv...	[ "Zeju Li", "Yizhou Zhou", "Qiang Xu" ]	[ "cs.LG", "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-01-31T00:00:00	https://arxiv.org/abs/2602.21221	https://arxiv.org/pdf/2602.21221v1	2602.21221	10.48550/arXiv.2602.21221	1	0	false	null	arXiv.org	0.4595
d804ea50b350643e4331b8dbe9a9d18bbd90dee658af47b9cc53c02386cc2b54	[ "arxiv", "semantic_scholar" ]	Epistemic Context Learning: Building Trust the Right Way in LLM-Based Multi-Agent Systems	Individual agents in multi-agent (MA) systems often lack robustness, tending to blindly conform to misleading peers. We show this weakness stems from both sycophancy and inadequate ability to evaluate peer reliability. To address this, we first formalize the learning problem of history-aware reference, introducing the ...	[ "Ruiwen Zhou", "Maojia Song", "Xiaobao Wu", "Sitao Cheng", "Xunjian Yin", "Yuxi Xie", "Zhuoqun Hao", "Wenyue Hua", "Liangming Pan", "Soujanya Poria", "Min-Yen Kan" ]	[ "cs.AI", "cs.CL", "cs.MA" ]	[ "Computer Science" ]	2026-01-29T00:00:00	https://arxiv.org/abs/2601.21742	https://arxiv.org/pdf/2601.21742v1	2601.21742	10.48550/arXiv.2601.21742	1	0	true	https://github.com/skyriver-2000/epistemic-context-learning	arXiv.org	0.7066
5c961e467799ace511d902bebffa8ef48bfc543c3ecc142fcc08bffeafaa805e	[ "arxiv", "semantic_scholar" ]	Self-Manager: Parallel Agent Loop for Long-form Deep Research	Long-form deep research requires multi-faceted investigations over extended horizons to get a comprehensive report. When handling such complex tasks, existing agents manage context at the subtask level to overcome linear context accumulation and information loss. However, they still adhere to a single context window an...	[ "Yilong Xu", "Zhi Zheng", "Xiang Long", "Yujun Cai", "Yiwei Wang" ]	[ "cs.CL", "cs.AI", "cs.IR" ]	[ "Computer Science" ]	2026-01-25T00:00:00	https://arxiv.org/abs/2601.17879	https://arxiv.org/pdf/2601.17879v1	2601.17879	10.48550/arXiv.2601.17879	1	0	false	null	arXiv.org	0.4526
7278e81ef9098d766f6d3acb7d2db59e9d11c200a2689c31cedf8960958d2031	[ "arxiv", "semantic_scholar" ]	Gated Sparse Attention: Combining Computational Efficiency with Training Stability for Long-Context Language Models	The computational burden of attention in long-context language models has motivated two largely independent lines of work: sparse attention mechanisms that reduce complexity by attending to selected tokens, and gated attention variants that improve training sta-bility while mitigating the attention sink phenomenon. We ...	[ "Alfred Shen", "Aaron Shen" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-01-12T00:00:00	https://arxiv.org/abs/2601.15305	https://arxiv.org/pdf/2601.15305v1	2601.15305	10.48550/arXiv.2601.15305	0	0	false	null	arXiv.org	0.4377
6f44d8298ad1ee1486089bec0800b1d87d5c5300504806d3cee5d79f92c33721	[ "arxiv", "semantic_scholar" ]	DYCP: Dynamic Context Pruning for Long-Form Dialogue with LLMs	Large Language Models (LLMs) increasingly operate over long-form dialogues with frequent topic shifts. While recent LLMs support extended context windows, efficient management of dialogue history in practice is needed due to inference cost and latency constraints. We present DyCP, a lightweight context management metho...	[ "Nayoung Choi", "Jonathan Zhang", "Jinho D. Choi" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-01-12T00:00:00	https://arxiv.org/abs/2601.07994	https://arxiv.org/pdf/2601.07994v5	2601.07994	10.48550/arXiv.2601.07994	1	0	false	null	arXiv.org	0.4377
0aab31a66da9c8bd3c4b129d9784f7c381c82292a016d37830bdca24e60a0565	[ "arxiv", "semantic_scholar" ]	Intelligence Degradation in Long-Context LLMs: Critical Threshold Determination via Natural Length Distribution Analysis	Large Language Models (LLMs) exhibit catastrophic performance degradation when processing contexts approaching certain critical thresholds, even when information remains relevant. This intelligence degradation-defined as over 30% drop in task performance-severely limits long-context applications. This degradation shows...	[ "Weiwei Wang", "Jiyong Min", "Weijie Zou" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-01-07T00:00:00	https://arxiv.org/abs/2601.15300	https://arxiv.org/pdf/2601.15300v1	2601.15300	10.48550/arXiv.2601.15300	5	0	true	null	arXiv.org	0.6676
78efcd52b71fc05b3c54ad8d4574e8e24308b057b4e76c1fc2b6f74d67ec3789	[ "arxiv", "semantic_scholar" ]	Not All Needles Are Found: How Fact Distribution and Don't Make It Up Prompts Shape Literal Extraction, Logical Inference, and Hallucination Risks in Long-Context LLMs	Large language models (LLMs) increasingly support very long input contexts. Yet it remains unclear how reliably they extract and infer information at scale. Performance varies with context length and strongly interacts with how information is distributed in real-world corpora. Motivated by these observations, we study ...	[ "Amirali Ebrahimzadeh", "Seyyed M. Salili" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-01-05T00:00:00	https://arxiv.org/abs/2601.02023	https://arxiv.org/pdf/2601.02023v1	2601.02023	10.48550/arXiv.2601.02023	0	0	false	null	arXiv.org	0.4297
243f02f2b418695987969db683ec8a2523536d6c403ccf8b41d7038890b528ea	[ "arxiv", "semantic_scholar" ]	Context as a Tool: Context Management for Long-Horizon SWE-Agents	Agents based on large language models have recently shown strong potential on real-world software engineering (SWE) tasks that require long-horizon interaction with repository-scale codebases. However, most existing agents rely on append-only context maintenance or passively triggered compression heuristics, which ofte...	[ "Shukai Liu", "Jian Yang", "Bo Jiang", "Yizhi Li", "Jinyang Guo", "Xianglong Liu", "Bryan Dai" ]	[ "cs.CL" ]	[ "Computer Science" ]	2025-12-26T00:00:00	https://arxiv.org/abs/2512.22087	https://arxiv.org/pdf/2512.22087v1	2512.22087	10.48550/arXiv.2512.22087	17	0	false	null	arXiv.org	0.4182
8394c56ad80db360672a9a08d121eb00f92b5184fdfd810feda4251e8d148cb3	[ "arxiv", "semantic_scholar" ]	Context Discipline and Performance Correlation: Analyzing LLM Performance and Quality Degradation Under Varying Context Lengths	The scaling trend in Large Language Models (LLMs) has prioritized increasing the maximum context window to facilitate complex, long-form reasoning and document analysis. However, managing this expanded context introduces severe computational overhead. This paper investigates the critical trade-off between system perfor...	[ "Ahilan Ayyachamy Nadar Ponnusamy", "Karthic Chandran", "M Maruf Hossain" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2025-12-25T00:00:00	https://arxiv.org/abs/2601.11564	https://arxiv.org/pdf/2601.11564v1	2601.11564	10.48550/arXiv.2601.11564	1	0	false	null	arXiv.org	0.4171
55203c81c15ca6a804e6ad22c4913450a531af5ac42c965c649bf55bbcd0c5d9	[ "arxiv", "semantic_scholar" ]	RePo: Language Models with Context Re-Positioning	In-context learning is fundamental to modern Large Language Models (LLMs); however, prevailing architectures impose a rigid and fixed contextual structure by assigning linear or constant positional indices. The rigid position information poses the full burden of organizing the input structure to attention layers, thus ...	[ "Huayang Li", "Tianyu Zhao", "Deng Cai", "Richard Sproat" ]	[ "cs.LG", "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2025-12-16T00:00:00	https://arxiv.org/abs/2512.14391	https://arxiv.org/pdf/2512.14391v3	2512.14391	10.48550/arXiv.2512.14391	2	1	true	https://github.com/SakanaAI/repo	arXiv.org	0.6286
18b8148da2d671c9551047bf58aa7c12e0082b6f14ade32ce03bc4f0a67be01e	[ "arxiv", "semantic_scholar" ]	Let's (not) just put things in Context: Test-Time Training for Long-Context LLMs	Progress on training and architecture strategies has enabled LLMs with millions of tokens in context length. However, empirical evidence suggests that such long-context LLMs can consume far more text than they can reliably use. On the other hand, it has been shown that inference-time compute can be used to scale perfor...	[ "Rachit Bansal", "Aston Zhang", "Rishabh Tiwari", "Lovish Madaan", "Sai Surya Duvvuri", "Devvrit Khatri", "David Brandfonbrener", "David Alvarez-Melis", "Prajjwal Bhargava", "Mihir Sanjay Kale", "Samy Jelassi" ]	[ "cs.LG", "cs.CL" ]	[ "Computer Science" ]	2025-12-15T00:00:00	https://arxiv.org/abs/2512.13898	https://arxiv.org/pdf/2512.13898v1	2512.13898	10.48550/arXiv.2512.13898	10	0	false	null	arXiv.org	0.4056
0c047ce00d9aad7cff9a60711a9e88631239d6af4e1152ff58ffcf7c0e3714b0	[ "arxiv", "semantic_scholar" ]	Extending the Context of Pretrained LLMs by Dropping Their Positional Embeddings	So far, expensive finetuning beyond the pretraining sequence length has been a requirement for effectively extending the context of language models (LM). In this work, we break this key bottleneck by Dropping the Positional Embeddings of LMs after training (DroPE). Our simple method is motivated by three key theoretica...	[ "Yoav Gelberg", "Koshi Eguchi", "Takuya Akiba", "Edoardo Cetin" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2025-12-13T00:00:00	https://arxiv.org/abs/2512.12167	https://arxiv.org/pdf/2512.12167v1	2512.12167	10.48550/arXiv.2512.12167	14	0	false	null	arXiv.org	0.4033
27d3ee1a30061fd942a4878c391d7378afe0edbfb830f25b07f3fde736306d2d	[ "arxiv", "semantic_scholar" ]	In-Context Learning for Seismic Data Processing	Seismic processing transforms raw data into subsurface images essential for geophysical applications. Traditional methods face challenges, such as noisy data, and manual parameter tuning, among others. Recently deep learning approaches have proposed alternative solutions to some of these problems. However, important ch...	[ "Fabian Fuchs", "Mario Ruben Fernandez", "Norman Ettrich", "Janis Keuper" ]	[ "cs.CV", "cs.LG" ]	[ "Computer Science" ]	2025-12-12T00:00:00	https://arxiv.org/abs/2512.11575	https://arxiv.org/pdf/2512.11575v2	2512.11575	10.48550/arXiv.2512.11575	0	0	true	null	arXiv.org	0.6216
dd55f26595320c6433dbd209cad47fcda5cd614787774afbd7003cfcada2bc4b	[ "arxiv", "semantic_scholar" ]	Learning to Extract Context for Context-Aware LLM Inference	User prompts to large language models (LLMs) are often ambiguous or under-specified, and subtle contextual cues shaped by user intentions, prior knowledge, and risk factors strongly influence what constitutes an appropriate response. Misinterpreting intent or risks may lead to unsafe outputs, while overly cautious inte...	[ "Minseon Kim", "Lucas Caccia", "Zhengyan Shi", "Matheus Pereira", "Marc-Alexandre Côté", "Xingdi Yuan", "Alessandro Sordoni" ]	[ "cs.LG" ]	[ "Computer Science" ]	2025-12-12T00:00:00	https://arxiv.org/abs/2512.11986	https://arxiv.org/pdf/2512.11986v1	2512.11986	10.48550/arXiv.2512.11986	1	0	false	null	arXiv.org	0.4022
79bf5fac5eb1656a2fb12cf366ec134ad0378c0319f064013abdeb26a40fde94	[ "arxiv", "semantic_scholar" ]	SWAA: Sliding Window Attention Adaptation for Efficient and Quality Preserving Long Context Processing	The quadratic complexity of self attention in Transformer based LLMs renders long context inference prohibitively expensive. While Sliding Window Attention (SWA), the simplest sparse attention pattern, offers a linear complexity alternative, it suffers from catastrophic long context performance collapse, which stems fr...	[ "Yijiong Yu", "Jiale Liu", "Qingyun Wu", "Huazheng Wang", "Ji Pei" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2025-12-11T00:00:00	https://arxiv.org/abs/2512.10411	https://arxiv.org/pdf/2512.10411v5	2512.10411	null	0	0	true	https://github.com/yuyijiong/sliding-window-attention-adaptation	null	0.474
2dc5cdca7a316bf7836b965959c874f48f66767090c22e7866fd99a4bfb5bcfc	[ "arxiv", "semantic_scholar" ]	Beyond Real: Imaginary Extension of Rotary Position Embeddings for Long-Context LLMs	Rotary Position Embeddings (RoPE) have become a standard for encoding sequence order in Large Language Models (LLMs) by applying rotations to query and key vectors in the complex plane. Standard implementations, however, utilize only the real component of the complex-valued dot product for attention score calculation. ...	[ "Xiaoran Liu", "Yuerong Song", "Zhigeng Liu", "Zengfeng Huang", "Qipeng Guo", "Zhaoxiang Liu", "Shiguo Lian", "Ziwei He", "Xipeng Qiu" ]	[ "cs.CL" ]	[ "Computer Science" ]	2025-12-08T00:00:00	https://arxiv.org/abs/2512.07525	https://arxiv.org/pdf/2512.07525v1	2512.07525	10.48550/arXiv.2512.07525	1	0	true	https://github.com/OpenMOSS/rope_pp	arXiv.org	0.6145

End of preview. Expand in Data Studio

Long-Context LLM Papers — FineSet

A research-paper dataset on Long-Context LLM Papers, assembled, deduplicated, and quality-scored by FineSet from arXiv and Semantic Scholar.

📸 This is a dated snapshot — generated 2026-06-19. It is not auto-updated. Research on Long-Context LLM Papers moves fast — new papers land on arXiv every week. Want this same dataset refreshed daily, on a topic you choose? See the bottom. ↓

Why this dataset

Quality-scored: quality_score float (0–1), blends citations with recency + code/venue signals — filter out the noise
Papers with code: 146 flagged via has_code — find reproducible work fast
Deduplicated: arXiv + Semantic Scholar cross-referenced, duplicate records merged
Clean JSONL: 437 records, one per line, normalized fields — no encoding garbage

Dataset details

Records: 437
Date range: 2022–2026
Snapshot date: 2026-06-19 (frozen — see note above)
Sources: arXiv, Semantic Scholar (cross-referenced, duplicates merged)
arXiv categories: cs.LG, cs.CL
Quality scoring: citations + recency + code/venue blend, 0–1 (p50=0.34, p90=0.615)
Format: JSONL, one record per line

Fields

Field	Type	Description
id	string	Deterministic SHA256 record id
sources	list	Which sources contributed (`arxiv`, `semantic_scholar`)
title	string	Paper title
abstract	string	Full abstract
authors	list	Author names
categories	list	arXiv category codes
fields_of_study	list	Semantic Scholar field tags
published_date	string	ISO 8601 date
url	string	arXiv abstract URL
pdf_url	string\|null	Open-access PDF if available
arxiv_id	string\|null	arXiv identifier
doi	string\|null	DOI if available
citation_count	int	Citation count (Semantic Scholar)
influential_citation_count	int	Influential citations (Semantic Scholar)
has_code	bool	Code repo detected in the arXiv comment
code_url	string\|null	GitHub URL if detected
venue	string\|null	Publication venue
quality_score	float	0–1, blended (citations + recency + code/venue)

Quality score methodology

quality_score = max(impact, freshness), clamped to [0, 1], where:

impact = max( log10(citations+1)/4 , log10(influential_citations+1)/2 ) — realized impact (0.5 at 100 citations, ~0.75 at 1,000, 1.0 at 10,000+).
freshness = recency × (0.35 + 0.30·has_code + 0.20·has_venue) — a baseline for recent papers (so a strong paper published this week isn't scored 0 just for lacking citations), where recency is 1.0 for papers ≤60 days old and decays linearly to 0 by ~18 months.

Old highly-cited papers score on impact; brand-new papers score on freshness; old uncited papers score ~0. Useful for filtering training data by quality, not just age.

👉 Want this on YOUR topic, updated daily?

This snapshot is frozen at 2026-06-19. The live FineSet pipeline keeps a dataset like this refreshed every day on whatever topic you describe — new papers in, dedup and quality scoring automatic, export as JSONL/Parquet or push straight to the Hub.

Tell me the topic you'd want and I'll run the pipeline on it — open a discussion on this dataset, it's free and it's how I decide what to build next.

→ fineset.io — describe what you want to train on, get a dataset. Early-access waitlist open (referral skip available).

Downloads last month: -