4687 papers
This paper introduces two software-only techniques, Overflow-Aware Scaling (OAS) and Macro Block Scaling (MBS), that significantly reduce the accuracy gap between the hardware-efficient MXFP4 format and NVIDIA's NVFP4 standard in Large Language Models, achieving near-parity performance with minimal computational overhead.
The paper introduces Design Conductor, an autonomous agent that leverages frontier models to independently design, verify, and generate a tape-out ready 1.48 GHz RISC-V CPU (VerCore) from a text specification to GDSII layout in just 12 hours, marking the first instance of an agent building a complete, working CPU end-to-end.
This paper introduces CktEvo, a repository-level benchmark and closed-loop framework that enables large language models to iteratively optimize Power, Performance, and Area (PPA) in complete RTL designs by preserving functional behavior across cross-file dependencies without human intervention.
The paper introduces SiliconMind-V1, a unified multi-agent framework that leverages testbench-driven verification and iterative debug-reasoning workflows to train locally fine-tuned LLMs for generating functionally correct Verilog RTL designs, outperforming state-of-the-art models with greater efficiency and privacy.
This paper presents ALADIN, an accuracy-latency-aware framework that enables the pre-deployment evaluation of mixed-precision quantized neural networks on scratchpad-based embedded AI accelerators by transforming models into platform-aware representations to analyze trade-offs and bottlenecks without requiring physical hardware.
This paper presents preliminary evidence from multi-agent simulations suggesting that alignment techniques and invisible censorship in large language models may paradoxically induce collective pathological behaviors and insight-action dissociation, indicating that safety interventions can sometimes cause the very harms they aim to prevent.
This PhD thesis presents novel, cost-efficient methods for assessing and enhancing the reliability of Deep Neural Network hardware accelerators, including a systematic literature review, new analytical tools, optimized trade-off methodologies, and the development of the AdAM real-time fault tolerance technique.
ARKV is a lightweight, adaptive framework that dynamically allocates precision levels to KV cache tokens based on per-layer attention dynamics and token importance, achieving a 4x reduction in memory usage while preserving ~97% of baseline accuracy for long-context LLM inference without requiring retraining or architectural modifications.
This cross-platform study evaluates blind reset as a measurement-free ancilla recycling technique on superconducting and trapped-ion processors, demonstrating that it can significantly reduce logical-cycle latency while maintaining high ancilla cleanliness and identifying specific architecture-dependent crossover points for optimal deployment.
This paper presents a systematic review and performance evaluation of Federated Learning in edge computing, benchmarking five leading algorithms across key metrics to identify trade-offs, highlight SCAFFOLD's superior accuracy and robustness versus FedAvg's efficiency, and propose a future research agenda to address challenges like data heterogeneity and energy limitations.
This paper introduces Auralink SDC, an edge-deployed multi-agent AI architecture that autonomously manages electric vehicle charging infrastructure with high reliability and sub-50ms latency, achieving 78% autonomous incident resolution and 87.6% diagnostic accuracy to address the critical failure rates and slow remediation times of current cloud-centric systems.
This paper presents a sensitivity-guided framework for compressing Reservoir Computing accelerators that systematically balances quantization and pruning to significantly improve hardware efficiency and reduce power consumption on FPGAs while maintaining high model accuracy across various time-series tasks.
This paper reviews FPGA-based AI accelerators for deep learning, highlighting their advantages over ASICs and GPUs, detailing key hardware optimization techniques such as loop pipelining and quantization, and analyzing state-of-the-art designs to identify challenges for future innovations.
The paper introduces Zipage, an LLM inference engine utilizing Compressed PagedAttention to combine token-wise KV cache eviction with PagedAttention, achieving over 2.1 speedup in high-concurrency reasoning tasks while maintaining approximately 95% of the performance of full KV inference.
This paper presents a systematic layer-wise and block-wise sensitivity analysis of NVFP4 and MXFP4 quantization across three Qwen2.5 model scales, revealing that MLP up- and down-projection layers are the most sensitive components while sensitivity patterns vary by format and model depth rather than being confined to final blocks.
This paper introduces the Variable-Invariant Two-Dimensional State Space Model (VI 2D SSM) and its unified VI 2D Mamba architecture, which theoretically establish and implement a permutation-equivariant framework for multivariate time series that eliminates artificial variable ordering to achieve state-of-the-art performance with improved structural scalability.
The paper introduces HCAPO, a novel framework that enhances long-horizon LLM agents by leveraging hindsight reasoning to refine step-level Q-values and employing a multi-scale advantage mechanism to address sparse reward challenges, thereby significantly outperforming state-of-the-art methods like GRPO on benchmarks such as WebShop and ALFWorld.
This paper introduces **Turn**, a compiled, actor-based programming language that enhances agentic software by integrating LLM inference as a typed primitive with schema validation, confidence-based control flow, isolated actor contexts, capability-based identity, and compile-time schema absorption to enforce critical safety and state invariants at the language level.
This paper introduces a principled method to reduce -invariant functions on product spaces to -invariant functions on alone, where is the isotropy subgroup of , thereby enabling flexible Equivariant Neural Fields to handle arbitrary group actions and heterogeneous product spaces without structural constraints.
The paper introduces EDMFormer, a transformer model trained on a newly released dataset of 98 professionally annotated EDM tracks (EDM-98) to address the limitations of existing music segmentation methods by leveraging genre-specific energy, rhythm, and timbre features for improved structure detection in Electronic Dance Music.