Quantization, distillation, pruning and serving work aimed at the same accuracy for less memory, latency and money.
Vision Transformers underperform convolutional networks when training data is scarce, and distilling convolutional inductive biases from a CNN teacher is an effective remedy that leaves the deployed model unchanged. General-purpose feature distillation, however, transfers little in this setting.
In large-scale pretraining, the algorithm, architecture, and systems decisions are conventionally made in disconnected stages. A scaling law stage selects an architecture and training recipe, optimizing loss under compute constraints, and a separate systems stage then optimizes the implementation for hardware efficiency.
Anchor-based pointwise LLM reranking scores each candidate against a shared reference passage to recover cross-document context at pointwise cost. We study when this actually helps, using GCCP/PAGC as a representative method.
\texttt{TensorSketch} by~\cite{pham2013fast,kar2012random} provides efficient sketching algorithms for high-dimensional polynomial kernels $\vec{x}^{\otimes p} \in \R^{d^p}$. \cite{kar2012random} uses dense Johnson-Lindenstrauss (JL)-type projections with computational cost $O(pDd)$, where $D$ denotes the sketch dimension, whereas~\cite{pham2013fast} extends the sparse \texttt{CountSketch}~\citep{count_sketch} algorithm, yielding a faster algorithm for high-dimensional sparse inputs with running time $O\big(p(\nnz{\vec{x}} + D \log D)\big)$.
Personalized Federated Reinforcement Learning (PFRL) takes a decentralized approach to storing and accessing information based on past experiences while keeping each client's data private during the learning of each client's policy. Many current methods for PFRL rely heavily on exploiting existing reinforcement learning reward signals to derive an optimal policy for each client, thereby neglecting exploration in non-stationary or sparse-reward environments.
Abundant visual information strengthens vision-language model (VLM) perception, yet massive visual tokens raise inference costs. Existing visual token pruning methods rely on similarity-based guidance, which exploits pairwise text-vision and vision-vision token correlations for compression.
arXiv:2608. 09447v1 Announce Type: cross Abstract: On-policy distillation (OPD) aligns a student with a teacher on trajectories sampled from the student itself, reducing the train-test state mismatch of offline distillation.
arXiv:2608. 07915v1 Announce Type: new Abstract: Large language models (LLMs) increasingly read long inputs in the agentic era, from whole documents and codebases to conversations across many turns.
arXiv:2608. 08224v1 Announce Type: new Abstract: Reinforcement learning post-training unlocks complex reasoning in LLMs.
arXiv:2608. 07974v1 Announce Type: new Abstract: Large language model (LLM) fine-tuning at the edge adapts the model to scenario-specific data while preserving privacy.
arXiv:2607. 04425v2 Announce Type: replace-cross Abstract: Recent advances in multimodal foundation models and agent systems have driven GUI agents from single-platform task execution toward cross-platform interaction.
arXiv:2608. 08684v1 Announce Type: cross Abstract: Long-context LLM inference is bottlenecked by KV cache memory, yet distributing a limited cache budget across layers remains challenging.
arXiv:2607. 28069v2 Announce Type: replace Abstract: Long-context retrieval and agentic workloads repeatedly reuse the same documents under changing instructions, histories, and document orders.
arXiv:2608. 09226v1 Announce Type: cross Abstract: Efficient text-to-image generation requires both reinforcement-learning (RL)-based reward alignment and few-step distillation, yet these procedures are typically performed sequentially, increasing training cost and risking the loss of reward gains during compression.
arXiv:2608. 09006v1 Announce Type: cross Abstract: Large Language Models (LLMs) have achieved remarkable success across a wide range of tasks.
arXiv:2608. 08525v1 Announce Type: new Abstract: Federated Learning is a distributed machine learning paradigm that trains a global model by aggregating local clients without sharing private data of each client.
arXiv:2608. 09233v1 Announce Type: new Abstract: Flow-matching models are now a mainstream method to image generation, but its adaptation to diverse downstream scenarios typically relies on post-training, which may cause conflicts among task-specific optimization objectives.
arXiv:2608. 07814v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) language models deliver high capacity at low per-token compute, but deploying them cheaply requires compressing their many expert weight matrices.
arXiv:2608. 08960v1 Announce Type: new Abstract: Multi-step language-model agents repeatedly process growing interaction histories, leading to substantial context costs.
arXiv:2608. 09595v1 Announce Type: new Abstract: Compressing large language models to two bits or fewer is increasingly feasible through block-wise post-training quantization; cross-block variants reconstruct neighboring Transformer blocks within a moving window.