Quantization, distillation, pruning and serving work aimed at the same accuracy for less memory, latency and money.
arXiv:2606. 02781v1 Announce Type: cross Abstract: Deep neural networks (DNNs) have achieved state-of-the-art performance across diverse domains.
arXiv:2606. 02608v1 Announce Type: new Abstract: We study a Marchenko--Pastur (MP) random-matrix approach to pruning deep neural networks with very small post-pruning fine-tuning budgets.
arXiv:2606. 02823v1 Announce Type: new Abstract: Two-bit weight quantization is attractive for memory-efficient LLM inference, but the standard W2 level set {-2,-1,0,+1} often collapses under aggressive W2A4/KV4 settings.
arXiv:2505. 24037v3 Announce Type: replace Abstract: Sparse large language models (LLMs) offer an attractive direction toward efficient deployment, but adapting them to downstream tasks remains challenging.
arXiv:2606. 02841v1 Announce Type: new Abstract: Deep neural networks learn representations where individual features often lack interpretable meaning; a single neuron may activate for scattered, unrelated inputs.
arXiv:2606. 02963v1 Announce Type: new Abstract: Production inference increasingly targets a heterogeneous mix of accelerators.
arXiv:2606. 03493v1 Announce Type: cross Abstract: Neural networks suffer from shortcut learning, where learned features generalize well to the training set but not to in-distribution (ID) or out-of-distribution (OOD) test sets.
arXiv:2605. 18629v2 Announce Type: replace Abstract: Sparse autoencoders (SAEs) are one of the main methods to interpret the inner workings of deep neural networks (DNNs), decomposing activations into higher-dimensional features.
arXiv:2606. 03532v1 Announce Type: cross Abstract: Self on-policy distillation trains a student policy against a teacher derived from its own parameter history, yet the teacher's update schedule -- which governs the \emph{temporal coupling} between teacher and student -- has not been systematically studied as a stability variable.
arXiv:2511. 17126v4 Announce Type: replace-cross Abstract: Emerging deep-learning-based lens library pre-training (LensLib-PT) pipeline offers a new avenue for blind lens aberration correction by training a universal neural network, demonstrating strong capability in handling diverse unknown optical degradations.
arXiv:2606. 03712v1 Announce Type: new Abstract: Graph Language Models (GLMs) have become a promising direction for adapting Large Language Models (LLMs) to graph learning tasks.
arXiv:2606. 03458v1 Announce Type: new Abstract: Test-time scaling is a powerful approach to obtain better reasoning in large language models, but it becomes memory-bottlenecked during long-horizon decoding, as the KV-cache grows.
arXiv:2606. 03940v1 Announce Type: cross Abstract: In robotics systems, vast amounts of visual data are easily captured at high resolution using low-cost, low-power hardware.
arXiv:2606. 03910v1 Announce Type: cross Abstract: Disaggregated LLM inference forces the KV cache to traverse the datacenter network before decoding begins, so transfer time enters directly into the Time to First Token (TTFT) budget.
arXiv:2605. 01712v2 Announce Type: replace Abstract: Pareto set learning (PSL) is an emerging paradigm in multi-objective optimization that trains neural networks to map preference vectors to Pareto optimal solutions.
arXiv:2606. 02951v1 Announce Type: cross Abstract: Deploying language-driven agents in robotics requires evaluations that reflect real-world task demands: natural-language instructions with reproducible outcomes.
arXiv:2606. 02673v1 Announce Type: new Abstract: Graphs have been used to enhance large language models (LLMs) for structured reasoning, mostly as external knowledge sources are provided to models at test time.
arXiv:2606. 02842v1 Announce Type: new Abstract: Multimodal spatial reasoning often relies on long chains of intermediate textual and visual thoughts, where accumulating visual tokens and dense cross-modal attention incur substantial computation and memory overhead.
arXiv:2505. 23725v3 Announce Type: replace Abstract: DiLoCo is a powerful framework for training large language models (LLMs), enabling larger optimal batch sizes and increased accelerator utilization under networking constraints.
arXiv:2606. 03939v1 Announce Type: cross Abstract: Federated Learning (FL) of foundation and edge models increasingly targets deployments where client data distributions drift over time, yet existing forgetting-mitigation methods assume each client's distribution is stationary.