A new arXiv paper (2610.10859v1) introduces a preference-driven unlearning framework that removes targeted concepts such as identity and NSFW content directly from few-step distilled (FSD) text-to-ima…
Researchers posted arXiv:2610.00061v1, introducing GAP-DPO (Geometry-Aligned Preference DPO), an iterative algorithm that selects preference pairs by aligning their gradient direction with expected us…
A practical guide details how to fine-tune agentic AI systems across four "dials" — training data, parameter-efficient fine-tuning, runtime hyperparameters, and preference alignment — using a support-…
Researchers introduced PRO-Step (arXiv:2609.01658v1), a training pipeline that uses a generative Process Reward Model (PRM) and step-level Direct Preference Optimization (DPO) to evaluate both logical…
MarkTechPost published a tutorial on August 20, 2026, detailing an end-to-end workflow for fine-tuning language models with Direct Preference Optimization (DPO), including auditing the Anthropic HH-RL…
Researchers propose inference-time mitigation strategies using Chain of Thought prompting and Direct Preference Optimization to shield large language models from adversarial political bias, raising Po…
A new video case study examines the transition of Direct Preference Optimization (DPO) from academic research to a frontier large language model, highlighting its impact on AI alignment. The presentat…
Natural language is the wrong tool for AI governance because autoregressive transformers cannot distinguish between actually checking a rule and generating a sentence that claims they checked it, acco…
A new study from arXiv finds that cross-lingual factual inconsistency in large language models can be mitigated at inference time, with persona prompting outperforming more invasive methods like Contr…
Researchers propose D2PO, a framework that optimizes diffusion sampling policies by reformulating sampler optimization as a preference-based alignment problem using Direct Preference Optimization (DPO…
Amazon Web Services introduced Reverse Direct Preference Optimization (rDPO), a novel unlearning technique behind Amazon Nova Customizable Content Moderation Settings (CCMS), which allows approved cus…
Researchers have developed a method called Emergent Alignment that enables large language models to self-correct unethical outputs by adding a conscience step and using Direct Preference Optimization.…
Researchers at arXiv propose a framework for selecting the most informative comparison pairs in preference-based post-training of large language models, showing that strategic pair selection can impro…
A comprehensive review published on arXiv examines fine-tuning techniques for Large Language Models (LLMs), covering methodologies from supervised and unsupervised learning to parameter-efficient meth…
Researchers led by Damani et al. introduced a method to train language models to express their uncertainty by adding a calibration reward to reinforcement learning from verifiable rewards (RLVR). The …
Researchers Liu et al. published a paper on arXiv 2025 introducing ProRL, a method using prolonged reinforcement learning with KL resets to expand reasoning boundaries in AI models. An interactive vis…
Zhao et al. published GenPRM, a generative process reward model that reasons and runs code to verify each step, achieving state-of-the-art performance where a 7B parameter model outperforms a 72B para…
Researchers have identified three systematic failure modes in Audio LLMs when transcribing English-Mandarin code-switching speech: language omission, translation-instead-of-transcription, and hallucin…