Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
MiniPIC removes the requirement for identical prefixes to reuse KV cache entries, enabling efficient caching of recurring structured inputs in retrieval-augmented and agentic workloads without the large server-side code changes or host-to-device transfer overhead of prior PIC approaches.
The framework and dataset directly extend multimodal medical AI to seven major Indian languages, addressing the lack of equitable AI-driven healthcare assistance in multilingual, low-resource settings like rural India that English-centric MLLMs cannot serve.
The work shows that a learned, cognitively grounded multi-factor value function substantially outperforms the recency and semantic-similarity heuristics currently used in production agent memory systems, and exposes a methodological flaw in how LongMemEval is commonly evaluated.
HarnessBridge replaces the manual engineering bottleneck in LLM agent harness design with an end-to-end trainable module, reducing token usage and trajectory length while maintaining competitive benchmark performance.
MSA demonstrates that a 109B-parameter model can process 1M-token contexts with 28.4x less attention compute and 14.2x faster prefill, making million-token agentic and code-reasoning workloads substantially more feasible at deployment scale.
TrajGenAgent demonstrates that a fine-tuning-free, hierarchical agent design can match or exceed the trajectory realism of computationally expensive fine-tuned models, lowering the barrier to generating privacy-safe synthetic mobility data for transportation, urban planning, and epidemic control applications.
AAA's single-interface design separates assessment logic from agent implementation, removing the heavy integration burden of existing LLM-centric harnesses and enabling reproducible, cross-agent comparisons that current fragmented benchmarks cannot support.
FrontierCode exposes a large gap between what current AI models can produce and what open-source maintainers would actually accept, with even the top-ranked model scoring only 13.4% on the hardest subset — a concrete signal that existing benchmarks have been overstating model readiness for production codebases.
TRACE directly addresses the repeated-friction failure mode where users must restate the same correction across sessions — a gap that memory-based approaches alone demonstrably fail to close.
AgentBuild shifts the durable artifact of scientific agent development from model-specific tuning to a scientist-authored contract, meaning workflow-scope failures become explicit contract failures and agent behavior can be re-tuned across model generations without a full rebuild.