arXiv Machine Learning
Aug 18
Conditional Evaluation of Language Models with Cheap Auxiliary Signals
arXiv:2608. 16210v1 Announce Type: new Abstract: Aggregate accuracy hides where models succeed and fail.
Predicting Human-Gain Curves under Local Proxy Optimization from Repeated Ratings
arXiv:2610. 41114v1 Announce Type: new Abstract: Optimizing a proxy objective can improve human-perceived quality, but the resulting gain can vary with optimization strength.
Related stories
arXiv Machine Learning
Sep 2
Post-Training Science for Supervised Fine-Tuning
arXiv:2609.01244v1 Announce Type: new Abstract: Every supervised fine-tuning run forces the same chain of decisions, such as learning rate, batch size, LoRA or full fine-tuning, how many epochs, whic...
arXiv AI
Jul 21
Normalized Rewards for Preference Optimization
arXiv:2607. 16240v1 Announce Type: cross Abstract: Direct Alignment Algorithms (DAAs) such as DPO have become a common way to post-train and align LLMs with human preferences.
Hugging Face Trending Papers
Jul 8
Online Data Selection Is Implicit Alignment
Supervised fine-tuning (SFT) is often treated as a capability-adaptation step, while alignment is attributed to later preference optimization or reinforcement learning. This separation is incomplete: when examples are scored and kept online during fine-tuning, the choice of which data to train on already changes the model's behavioral preferences.
arXiv Machine Learning
Aug 4
UpliftBench: Revealing Outcome-Regime and Objective Mismatch in Uplift Evaluation
arXiv:2608. 00915v1 Announce Type: new Abstract: Uplift modeling (conditional-average-treatment-effect estimation) drives personalized targeting, yet published uplift benchmarks frequently disagree on which estimator performs best; we show the disagreement is substantially about metrics, not models.
arXiv Machine Learning
Jul 9
Online Data Selection Is Implicit Alignment
arXiv:2607. 07023v1 Announce Type: new Abstract: Supervised fine-tuning (SFT) is often treated as a capability-adaptation step, while alignment is attributed to later preference optimization or reinforcement learning.
Hugging Face Trending Papers
Sep 1
Post-Training Science for Supervised Fine-Tuning
Every supervised fine-tuning run forces the same chain of decisions, such as learning rate, batch size, LoRA or full fine-tuning, how many epochs, which optimiser, and what data to feed the model. Eac...
arXiv Machine Learning
Oct 3
MECHVAR: Variance-Guided Mechanism Discrimination for Autonomous Machine Learning Experiment Selection
MECHVAR is a lightweight, auditable rule for selecting experiments from a finite library to discriminate between candidate mechanisms. It chooses probes by maximizing the posterior‑weighted variance of predicted responses, a score that aligns with the Box–Hill pairwise‑KL criterion and links to expected information gain when separations are small. Experiments on a 25‑block audit and a Digits loop show MECHVAR outperforming confirmation‑first strategies and matching or exceeding EIG in identification accuracy while being far faster to compute.
arXiv AI
Oct 21
The Reward Model Selection Crisis in Personalized Alignment
arXiv:2512.23067v3 Announce Type: replace Abstract: Personalized alignment from preference data has focused primarily on improving personal reward model (RM) accuracy, with the implicit assumption th...
arXiv AI
Jun 29
When Is an LLM Worth It for Hyperparameter Optimization? A Budget-Matched Study on Tabular Data Finds the Warm-Start Is a Default Configuration, Not the Model
arXiv:2606. 21641v2 Announce Type: replace-cross Abstract: Large language models (LLMs) have been proposed as hyperparameter-optimization (HPO) advisors that "warm-start" search from prior knowledge, proposing strong configurations in very few evaluations.
arXiv AI
Jun 2
Reward Score Matching: Unifying Reward-based Fine-tuning for Flow and Diffusion Models
arXiv:2604. 17415v3 Announce Type: replace-cross Abstract: Reward-based fine-tuning steers a pretrained diffusion or flow-based generative model toward higher-reward samples while remaining close to the pretrained model.
arXiv Machine Learning
Sep 24
tidyHEBO: Robust General-Purpose Bayesian Optimization with Model-Consistent Warping and Pareto Search
tidyHEBO is a BoTorch-native Bayesian optimization tool that jointly applies Yeo-Johnson output warping to a Gaussian‑process surrogate, evaluates acquisition functions on the original objective scale, and conducts constrained cumulative Pareto search across multiple acquisition criteria. Using only default settings, it outperformed other methods on the Olympus benchmark and performed strongly on synthetic, Needle‑in‑a‑Haystack, and Bayesmark tasks, while adaptive batching offered a trade‑off between parallelization and optimization quality. These results position tidyHEBO as a robust, reproducible optimizer suitable for diverse practical problems, including scientific applications and hyperparameter tuning.