ALU uses public data to suppress unlearning cost quadratically while characterizing distribution mismatch effects, enabling mass unlearning with maintained utility.
hub
and Daly, Raymond E
13 Pith papers cite this work. Polarity classification is still indexing.
13
Pith papers citing it
hub tools
representative citing papers
Shapley values in product games equal the integral of a degree-(d-1) polynomial over [0,1], allowing provably exact or near-exact computation via Gauss-Legendre quadrature with O(d m_q) work.
mPL measures attacker-aligned privacy leakage from joint data releases and AmPL provides an adaptive way to bound it with low utility cost in ML settings.
Min-K% Prob detects pretraining data in LLMs by flagging outlier low-probability words in text, achieving 7.4% better performance than prior methods on the new WIKIMIA benchmark.
Activation Addition steers language models by adding contrastive activation vectors from prompt pairs to control high-level properties like sentiment and toxicity at inference time without training.
Longformer uses local windowed attention plus task-specific global attention to achieve linear scaling and state-of-the-art results on long-document language modeling, QA, and summarization after pretraining.
A convolution-smoothed and debiased SVM estimator achieves asymptotic normality for statistical inference in high-dimensional offline and online settings.
Summing outputs from separately trained QLoRA PEFT modules provides strong performance for attribute-controlled text generation, often matching or exceeding single-task modules even on single-attribute tests.
E-value sequential tests enable early stopping of MCMC sampling in Bayesian deep ensembles, often needing only a fraction of the full budget while improving over standard deep ensembles.
Agent Q integrates MCTS-guided search, self-critique, and off-policy DPO to train LLM agents that outperform behavior cloning and reinforced fine-tuning baselines in WebShop and achieve up to 95.4% success in real-world booking scenarios.
NV-Embed achieves first place on the MTEB leaderboard across 56 tasks by combining a latent attention layer, causal-mask removal, two-stage contrastive training, and data curation for LLM-based embedding models.
POVID generates AI-created preference data to fine-tune vision-language models with DPO, reducing hallucinations and improving benchmark scores.
RAFT aligns generative models by ranking samples with a reward model and fine-tuning only on the top-ranked outputs, reporting gains on reward scores and automated metrics for LLMs and diffusion models.
citing papers explorer
-
Detecting Pretraining Data from Large Language Models
Min-K% Prob detects pretraining data in LLMs by flagging outlier low-probability words in text, achieving 7.4% better performance than prior methods on the new WIKIMIA benchmark.
-
Steering Language Models With Activation Engineering
Activation Addition steers language models by adding contrastive activation vectors from prompt pairs to control high-level properties like sentiment and toxicity at inference time without training.
-
RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment
RAFT aligns generative models by ranking samples with a reward model and fine-tuning only on the top-ranked outputs, reporting gains on reward scores and automated metrics for LLMs and diffusion models.