LACUNA is a new testbed that injects PII into predefined model parameters to benchmark the localization precision of LLM unlearning methods, revealing that SOTA approaches are imprecise despite strong output performance.
hub
Thirty-seventh Conference on Neural Information Processing Systems , year=
15 Pith papers cite this work. Polarity classification is still indexing.
15
Pith papers citing it
hub tools
representative citing papers
WildChat releases a dataset of 1 million ChatGPT conversations with timestamps, demographics, and headers, claimed to be the most diverse and multilingual such resource available.
PARAMΔ upcycles dense models to MoE for per-language experts and grafts post-training deltas to enable data-efficient language expansion while preserving original capabilities.
Magpie synthesizes 300K high-quality alignment instructions from Llama-3-Instruct via auto-regressive prompting on partial templates, enabling fine-tuned models to match official instruct performance on AlpacaEval, ArenaHard, and WildBench.
Iterative self-rewarding via LLM-as-Judge in DPO training on Llama 2 70B improves instruction following and self-evaluation, outperforming GPT-4 on AlpacaEval 2.0.
LLMs rarely recognize their own prefilled (adversarially forced) outputs, and the recognition they do show tracks refusal circuits and question phrasing rather than a robust sense of self.
DDE introduces a compact coordinator network that combines denoised outputs from pre-trained diffusion models to enable generation in larger domains and complex conditioning settings.
GUARD-IT removes forget-set influence from LLMs via training-free, gated residual-stream rotations at inference, matching gradient unlearning baselines without weight edits.
Fine-tuning an LLM with DPO against a detector score plus a task metric (PUPPET) improves both the detector's AUROC and downstream task scores, beating watermarked baselines on task quality in the reported benchmarks.
ETW uses predictive entropy as a proxy for token informativeness to improve selective unlearning in LLMs, achieving better forgetting with less utility loss than prior token-level methods.
The paper maps agent memory research via three forms (token-level, parametric, latent), three functions (factual, experiential, working), and dynamics of formation/evolution/retrieval, plus benchmarks and future directions.
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.
IAP uses RL to train LLMs to explicitly infer and apply implicit user intent in single-turn personalized QA, achieving ~7.5% average macro-score gains over baselines on LaMP-QA.
POVID generates AI-created preference data to fine-tune vision-language models with DPO, reducing hallucinations and improving benchmark scores.
citing papers explorer
-
LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning
LACUNA is a new testbed that injects PII into predefined model parameters to benchmark the localization precision of LLM unlearning methods, revealing that SOTA approaches are imprecise despite strong output performance.
-
WildChat: 1M ChatGPT Interaction Logs in the Wild
WildChat releases a dataset of 1 million ChatGPT conversations with timestamps, demographics, and headers, claimed to be the most diverse and multilingual such resource available.
-
A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$\Delta$ Integration into Upcycled MoE
PARAMΔ upcycles dense models to MoE for per-language experts and grafts post-training deltas to enable data-efficient language expansion while preserving original capabilities.
-
Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing
Magpie synthesizes 300K high-quality alignment instructions from Llama-3-Instruct via auto-regressive prompting on partial templates, enabling fine-tuned models to match official instruct performance on AlpacaEval, ArenaHard, and WildBench.
-
Self-Rewarding Language Models
Iterative self-rewarding via LLM-as-Judge in DPO training on Llama 2 70B improves instruction following and self-evaluation, outperforming GPT-4 on AlpacaEval 2.0.
-
Can LLMs Reliably Self-Report Adversarial Prefills, and How?
LLMs rarely recognize their own prefilled (adversarially forced) outputs, and the recognition they do show tracks refusal circuits and question phrasing rather than a robust sense of self.
-
Diffusion Domain Expansion: Learning to Coordinate Pre-trained Diffusion Models
DDE introduces a compact coordinator network that combines denoised outputs from pre-trained diffusion models to enable generation in larger domains and complex conditioning settings.
-
Inference-Time Machine Unlearning via Gated Activation Redirection
GUARD-IT removes forget-set influence from LLMs via training-free, gated residual-stream rotations at inference, matching gradient unlearning baselines without weight edits.
-
LLM Output Detectability and Task Performance Can be Jointly Optimized
Fine-tuning an LLM with DPO against a detector score plus a task metric (PUPPET) improves both the detector's AUROC and downstream task scores, beating watermarked baselines on task quality in the reported benchmarks.
-
Forget What Matters, Keep the Rest: Selective Unlearning of Informative Tokens
ETW uses predictive entropy as a proxy for token informativeness to improve selective unlearning in LLMs, achieving better forgetting with less utility loss than prior token-level methods.
-
Memory in the Age of AI Agents
The paper maps agent memory research via three forms (token-level, parametric, latent), three functions (factual, experiential, working), and dynamics of formation/evolution/retrieval, plus benchmarks and future directions.
-
Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents
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.
-
Training LLMs with Reinforcement Learning for Intent-Aware Personalized Question Answering
IAP uses RL to train LLMs to explicitly infer and apply implicit user intent in single-turn personalized QA, achieving ~7.5% average macro-score gains over baselines on LaMP-QA.
-
Aligning Modalities in Vision Large Language Models via Preference Fine-tuning
POVID generates AI-created preference data to fine-tune vision-language models with DPO, reducing hallucinations and improving benchmark scores.
- Hybrid Policy Distillation for LLMs