MemTrace shows that evidence utilization, not retrieval, is the dominant failure mode in LLM long-term memory systems across tested configurations.
hub
∞-bench: Extending long context evaluation beyond 100k tokens
19 Pith papers cite this work, alongside 2 external citations. Polarity classification is still indexing.
19
Pith papers citing it
2
external citations · Pith
abstract
Processing and reasoning over long contexts is crucial for many practical applications of Large Language Models (LLMs), such as document comprehension and agent construction. Despite recent strides in making LLMs process contexts with more than 100K tokens, there is currently a lack of a standardized benchmark to evaluate this long-context capability. Existing public benchmarks typically focus on contexts around 10K tokens, limiting the assessment and comparison of LLMs in processing longer contexts. In this paper, we propose $\infty$Bench, the first LLM benchmark featuring an average data length surpassing 100K tokens. $\infty$Bench comprises synthetic and realistic tasks spanning diverse domains, presented in both English and Chinese. The tasks in $\infty$Bench are designed to require well understanding of long dependencies in contexts, and make simply retrieving a limited number of passages from contexts not sufficient for these tasks. In our experiments, based on $\infty$Bench, we evaluate the state-of-the-art proprietary and open-source LLMs tailored for processing long contexts. The results indicate that existing long context LLMs still require significant advancements to effectively process 100K+ context. We further present three intriguing analyses regarding the behavior of LLMs processing long context.
hub tools
citation-role summary
background 3
dataset 1
citation-polarity summary
representative citing papers
Audits reveal no reasoning benchmark controls position/filler/length jointly; CRE shows LLMs drop up to 88pp on middle-position tasks at 64K context, with diagnostic probe supporting positional cause.
LongMemEval-V2 is a new benchmark where AgentRunbook-C reaches 72.5% accuracy on long-term agent memory tasks, beating RAG baselines at 48.5% and basic coding agents at 69.3%.
SnapStream deploys sparse KV attention in a production inference system on dataflow accelerators, delivering 4x on-chip memory savings for DeepSeek-671B at 128k context with up to 1832 tokens/sec and minimal accuracy loss on LongBench-v2, AIME24, and LiveCodeBench.
KV cache compression causes task-dependent degradation in high-density reasoning due to disrupted CoT links; ShotKV mitigates this by preserving few-shot examples as indivisible semantic units through phase separation, delivering 9-18% accuracy gains and 11% latency reduction.
MLVU is a new benchmark for long video understanding that uses extended videos across diverse genres and multi-task evaluations, revealing that current MLLMs struggle significantly and degrade sharply with longer durations.
KV-cache eviction, prompt compression, recurrent state bounding, and agent memory consolidation are unified as one rate-distortion problem with a shared lower bound, shared failure mode, and transferable mechanisms.
HPP decouples perception from reasoning in long-video VLMs by having an LLM run iterative programmatic probes on hierarchically segmented video, reporting gains on LongVideoBench, EgoSchema, VideoMME, and MLVU.
NarrativeWorldBench evaluates 21 LLMs on nine narrative metrics across horizons to 200 episodes and introduces N-VSSM, a 256-dimensional variational state-space model that achieves plot-beat F1 >=0.84 with 4x lower compute and wins writer preference on consistency.
Lexical density acts as an independent limiter on effective LLM context windows, with performance collapsing from near-perfect to below 60% as information density rises in controlled ~12k-token benchmarks.
LaProx reformulates KV cache eviction as an output-aware matrix approximation, enabling a unified global token selection strategy that preserves LLM performance at 5% cache size across long-context benchmarks.
CL-bench Life shows frontier language models achieve only 13.8% average success on real-life context tasks, with the best model at 19.3%.
MemSearch-o1 mitigates memory dilution in agentic LLM search through reasoning-aligned token-level memory growth, retracing with a contribution function, and path reorganization, improving reasoning activation on benchmarks.
MSA is an end-to-end trainable memory model using sparse attention and document-wise RoPE that scales to 100M tokens with linear complexity and less than 9% degradation.
MiniCPM 1.2B and 2.4B models reach parity with 7B-13B LLMs via model wind-tunnel scaling and a WSD scheduler that yields a higher optimal data-to-model ratio than Chinchilla scaling.
Organizes the heterogeneous LLM prefill-decode design space along four axes and extracts three boundary decisions with guidance on precision, KV representation, and ownership.
Presents PopMedQA benchmark and shows domain-independent LLM methods fail on token-inefficient longitudinal medical records, leaving room for domain-specific approaches.
AgentCE-Bench is a lightweight grid-planning benchmark that controls task horizon via hidden slots H and difficulty via decoy budget B, validated across 13 models for consistent and discriminative evaluation.
StateX post-trains RNNs to expand recurrent state size, improving recall and in-context learning with negligible parameter growth.
citing papers explorer
-
MemTrace: Probing What Final Accuracy Misses in Long-Term Memory
MemTrace shows that evidence utilization, not retrieval, is the dominant failure mode in LLM long-term memory systems across tested configurations.
-
Positional Failures in Long-Context LLMs: A Blind Spot in Reasoning Benchmarks
Audits reveal no reasoning benchmark controls position/filler/length jointly; CRE shows LLMs drop up to 88pp on middle-position tasks at 64K context, with diagnostic probe supporting positional cause.
-
LongMemEval-V2: Evaluating Long-Term Agent Memory Toward Experienced Colleagues
LongMemEval-V2 is a new benchmark where AgentRunbook-C reaches 72.5% accuracy on long-term agent memory tasks, beating RAG baselines at 48.5% and basic coding agents at 69.3%.
-
SnapStream: Efficient Long Sequence Decoding on Dataflow Accelerators
SnapStream deploys sparse KV attention in a production inference system on dataflow accelerators, delivering 4x on-chip memory savings for DeepSeek-671B at 128k context with up to 1832 tokens/sec and minimal accuracy loss on LongBench-v2, AIME24, and LiveCodeBench.
-
Semantic Integrity Matters: Benchmarking and Preserving High-Density Reasoning in KV Cache Compression
KV cache compression causes task-dependent degradation in high-density reasoning due to disrupted CoT links; ShotKV mitigates this by preserving few-shot examples as indivisible semantic units through phase separation, delivering 9-18% accuracy gains and 11% latency reduction.
-
MLVU: Benchmarking Multi-task Long Video Understanding
MLVU is a new benchmark for long video understanding that uses extended videos across diverse genres and multi-task evaluations, revealing that current MLLMs struggle significantly and degrade sharply with longer durations.
-
What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents
KV-cache eviction, prompt compression, recurrent state bounding, and agent memory consolidation are unified as one rate-distortion problem with a shared lower bound, shared failure mode, and transferable mechanisms.
-
HPP: Hierarchical Programmatic Probing for Long Video Understanding by Decoupling Perception and Reasoning
HPP decouples perception from reasoning in long-video VLMs by having an LLM run iterative programmatic probes on hierarchically segmented video, reporting gains on LongVideoBench, EgoSchema, VideoMME, and MLVU.
-
NarrativeWorldBench: A Frontier-Saturated Benchmark and a Latent World Model for Long-Horizon Co-Creative Audio Drama
NarrativeWorldBench evaluates 21 LLMs on nine narrative metrics across horizons to 200 episodes and introduces N-VSSM, a 256-dimensional variational state-space model that achieves plot-beat F1 >=0.84 with 4x lower compute and wins writer preference on consistency.
-
Dense Contexts Are Hard Contexts: Lexical Density Limits Effective Context in LLMs
Lexical density acts as an independent limiter on effective LLM context windows, with performance collapsing from near-perfect to below 60% as information density rises in controlled ~12k-token benchmarks.
-
Reformulating KV Cache Eviction Problem for Long-Context LLM Inference
LaProx reformulates KV cache eviction as an output-aware matrix approximation, enabling a unified global token selection strategy that preserves LLM performance at 5% cache size across long-context benchmarks.
-
CL-bench Life: Can Language Models Learn from Real-Life Context?
CL-bench Life shows frontier language models achieve only 13.8% average success on real-life context tasks, with the best model at 19.3%.
-
MemSearch-o1: Empowering Large Language Models with Reasoning-Aligned Memory Growth in Agentic Search
MemSearch-o1 mitigates memory dilution in agentic LLM search through reasoning-aligned token-level memory growth, retracing with a contribution function, and path reorganization, improving reasoning activation on benchmarks.
-
MSA: Memory Sparse Attention for Efficient End-to-End Memory Model Scaling to 100M Tokens
MSA is an end-to-end trainable memory model using sparse attention and document-wise RoPE that scales to 100M tokens with linear complexity and less than 9% degradation.
-
MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies
MiniCPM 1.2B and 2.4B models reach parity with 7B-13B LLMs via model wind-tunnel scaling and a WSD scheduler that yields a higher optimal data-to-model ratio than Chinchilla scaling.
-
Demystifying the Design Space and Best Practices for Heterogeneous LLM Inference and Serving
Organizes the heterogeneous LLM prefill-decode design space along four axes and extracts three boundary decisions with guidance on precision, KV representation, and ownership.
-
The Verbose Context Problem in Medical Records
Presents PopMedQA benchmark and shows domain-independent LLM methods fail on token-inefficient longitudinal medical records, leaving room for domain-specific approaches.
-
AgentCE-Bench: Agent Configurable Evaluation with Scalable Horizons and Controllable Difficulty under Lightweight Environments
AgentCE-Bench is a lightweight grid-planning benchmark that controls task horizon via hidden slots H and difficulty via decoy budget B, validated across 13 models for consistent and discriminative evaluation.
-
StateX: Enhancing RNN Recall via Post-training State Expansion
StateX post-trains RNNs to expand recurrent state size, improving recall and in-context learning with negligible parameter growth.