LCDD creates sparse carriers for SFT behaviors that SFT-Eraser can reverse, with ablations showing the sparse structure enables causal control.
hub Mixed citations
Judging llm-as-a-judge with mt-bench and chatbot arena.Advances in neural information processing systems, 36:46595–46623
Mixed citation behavior. Most common role is background (29%).
35
Pith papers citing it
Background
29% of classified citations
hub tools
citation-role summary
background 2
dataset 2
method 2
baseline 1
citation-polarity summary
years
2026 35representative citing papers
VoxSafeBench reveals that speech language models recognize social norms from text but fail to apply them when acoustic cues like speaker or scene determine the appropriate response.
VGenST-Bench is a new video benchmark for MLLM spatio-temporal reasoning built via generative synthesis, a multi-agent pipeline with human oversight, a 3x2x2 taxonomy, and hierarchical tasks separating perception from reasoning.
LASH adaptively composes multiple jailbreak seed prompts via genetic search over subsets and mixture weights to reach 84.5% keyword ASR and 74.5% two-stage ASR on JailbreakBench while using only 30 queries per prompt.
CAPS is a four-stage inference-only cascade that adapts how much of each solution the verifier sees and how comparisons are distributed, halving per-candidate verifier tokens while outperforming uniform pairwise verification on most benchmarks.
AgentForesight introduces an online auditor model that predicts decisive errors in multi-agent trajectories at the earliest step using a coarse-to-fine reinforcement learning recipe on a new curated dataset AFTraj-2K.
A single safety demonstration appended at inference time mitigates many-shot jailbreak attacks by counteracting implicit malicious fine-tuning on harmful examples.
DAPRO provides the first dynamic, theoretically guaranteed way to allocate interaction budgets across test cases for bounding time-to-event in multi-turn LLM evaluations, achieving tighter coverage than static conformal survival methods.
Weblica scales RL training for visual web agents by building thousands of reproducible environments through HTTP caching for stable replays and LLM synthesis from real sites, yielding an 8B model that beats similar open baselines on navigation benchmarks.
XL-SafetyBench is a new cross-cultural benchmark showing frontier LLMs decouple jailbreak robustness from cultural sensitivity while local models trade off attack success against neutral-safe rates in a near-linear pattern indicating generation failure rather than alignment.
Green Shielding introduces CUE criteria and the HCM-Dx benchmark to demonstrate that routine prompt variations systematically alter LLM diagnostic behavior along clinically relevant dimensions, producing Pareto-like tradeoffs in plausibility versus coverage.
This paper introduces a systems-level conceptual framing and a three-level taxonomy (intra-model, system-level, socio-technical) for uncertainty propagation in compound LLM applications, along with engineering insights and open challenges.
KVBench reveals major gaps in current T2I models for knowledge-intensive tasks, and KE-Check narrows the gap between open- and closed-source models by adding structured knowledge and enforcing constraints.
LLM planners for robots often produce dangerous plans even when planning succeeds, with safety awareness staying flat as model scale improves planning ability.
R2ABench benchmark shows LLMs generate syntactically valid software architectures from requirements but produce structurally fragmented results due to weak relational reasoning.
This work delivers the first measurements of performance-energy trade-offs across four multi-request LLM workflow patterns on A100 GPUs using vLLM and Parrot.
ChunkFT enables full-parameter fine-tuning of Llama 3-8B on one 24 GB GPU and Llama 3-70B on two 80 GB GPUs by streaming gradients over dynamically activated sub-tensors.
GPRL carries a k-dimensional skew-symmetric preference structure into policy updates with per-dimension advantages and a drift monitor, yielding 56.51% length-controlled win rate on AlpacaEval 2.0 from Llama-3-8B-Instruct while outperforming SimPO and SPPO on other benchmarks.
Attention Hijacking is a new attack that improves cross-query transferability in VLMs by explicitly steering internal attention to a persistent image-dominant pattern.
PPOW uses window-level RL with cost-aware speedup and proximity rewards plus adaptive divergence-aware windowing to reach 6.29-6.52 acceptance lengths and 3.39-4.36x speedups in speculative decoding.
Bell-shaped time sampling accelerates masked diffusion language model training by roughly 4x on LM1B by countering locality bias in language data.
DeltaRubric decomposes multimodal preference evaluation into self-generated planning and verification steps within a single model, producing large accuracy improvements on VL-RewardBench via multi-role reinforcement learning.
S2FT replaces the sparse-spectrum assumption of prior Fourier PEFT with a learned rearrangement that maps a pre-estimated weight change into a domain where few spectral coefficients suffice.
LLM safety judges flip verdicts on equivalent policy rewrites up to 9.1% of the time and cannot distinguish meaningful from meaningless changes, requiring new invariance-based reliability metrics.
citing papers explorer
-
Crafting Reversible SFT Behaviors in Large Language Models
LCDD creates sparse carriers for SFT behaviors that SFT-Eraser can reverse, with ablations showing the sparse structure enables causal control.
-
How Many Iterations to Jailbreak? Dynamic Budget Allocation for Multi-Turn LLM Evaluation
DAPRO provides the first dynamic, theoretically guaranteed way to allocate interaction budgets across test cases for bounding time-to-event in multi-turn LLM evaluations, achieving tighter coverage than static conformal survival methods.
-
ChunkFT: Byte-Streamed Optimization for Memory-Efficient Full Fine-Tuning
ChunkFT enables full-parameter fine-tuning of Llama 3-8B on one 24 GB GPU and Llama 3-70B on two 80 GB GPUs by streaming gradients over dynamically activated sub-tensors.
-
General Preference Reinforcement Learning
GPRL carries a k-dimensional skew-symmetric preference structure into policy updates with per-dimension advantages and a drift monitor, yielding 56.51% length-controlled win rate on AlpacaEval 2.0 from Llama-3-8B-Instruct while outperforming SimPO and SPPO on other benchmarks.
-
Understanding and Accelerating the Training of Masked Diffusion Language Models
Bell-shaped time sampling accelerates masked diffusion language model training by roughly 4x on LM1B by countering locality bias in language data.
-
Fine-Tuning Without Forgetting via Loss-Adaptive Learning Rates
FINCH is a loss-adaptive learning-rate schedule that reduces forgetting by 93% on average during LLM fine-tuning while matching standard task performance across several benchmarks.
-
Guardian-as-an-Advisor: Advancing Next-Generation Guardian Models for Trustworthy LLMs
Guardian-as-an-Advisor prepends risk labels and explanations from a guardian model to queries, improving LLM safety compliance and reducing over-refusal while adding minimal compute overhead.