Large language models (LLMs) have achieved remarkable performance across many generation tasks. Nevertheless, effectively aligning them with desired behaviors remains a significant challenge. Activation steering is an effective and cost-efficient approach that directly modifies the activations of LLMs during the inference stage, aligning their responses with the desired behaviors and avoiding the high cost of fine-tuning. Existing methods typically indiscriminately intervene to all generations or rely solely on the question to determine intervention, which limits the accurate assessment of the intervention strength. To this end, we propose the Flexible Activation Steering with Backtracking (FASB) framework, which dynamically determines both the necessity and strength of intervention by tracking the internal states of the LLMs during generation, considering both the question and the generated content. Since intervening after detecting a deviation from the desired behavior is often too late, we further propose the backtracking mechanism to correct the deviated tokens and steer the LLMs toward the desired behavior. Extensive experiments on the TruthfulQA dataset and six multiple-choice datasets demonstrate that our method outperforms baselines. Our code will be released at https://github.com/gjw185/FASB.

The preservation of privacy has emerged as a critical topic in the era of artificial intelligence. However, current work focuses on user-oriented privacy, overlooking severe enterprise data leakage risks exacerbated by the Retrieval-Augmented Generation paradigm. To address this gap, our paper introduces a novel objective: enterprise-oriented privacy concerns. Achieving this objective requires overcoming two fundamental challenges: existing methods such as data sanitization severely degrade model performance, and the field lacks public datasets for evaluation. We address these challenges with several solutions. (1) To prevent performance degradation, we propose ABack, a training-free mechanism that leverages a Hidden State Model to pinpoint the origin of a leakage intention and rewrite the output safely. (2) To solve the lack of datasets, we construct PriGenQA, a new benchmark for enterprise privacy scenarios in healthcare and finance. To ensure a rigorous evaluation, we move beyond simple static attacks by developing a powerful adaptive attacker with Group Relative Policy Optimization. Experiments show that against this superior adversary, ABack improves the overall privacy utility score by up to 15\% over strong baselines, avoiding the performance trade-offs of prior methods.

This article presents an experiment focused on optimizing the MLOps (Machine Learning Operations) process, a crucial aspect of efficiently implementing machine learning projects. The objective is to identify patterns and insights to enhance the MLOps workflow, considering its iterative and interdependent nature in real-world model development scenarios. The experiment involves a comprehensive MLOps workflow, covering essential phases like problem definition, data acquisition, data preparation, model development, model deployment, monitoring, management, scalability, and governance and compliance. Practical tips and recommendations are derived from the results, emphasizing proactive planning and continuous improvement for the MLOps workflow. The experimental investigation was strategically integrated within a real-world ML project which followed essential phases of the MLOps process in a production environment, handling large-scale structured data. A systematic tracking approach was employed to document revisits to specific phases from a main phase under focus, capturing the reasons for such revisits. By constructing a matrix to quantify the degree of overlap between phases, the study unveils the dynamic and iterative nature of the MLOps workflow. The resulting data provides visual representations of the MLOps process's interdependencies and iterative characteristics within the experimental framework, offering valuable insights for optimizing the workflow and making informed decisions in real-world scenarios. This analysis contributes to enhancing the efficiency and effectiveness of machine learning projects through an improved MLOps process.

In a recent joint work, the author has developed a modification of Newton's method, named New Q-Newton's method, which can avoid saddle points and has quadratic rate of convergence. While good theoretical convergence guarantee has not been established for this method, experiments on small scale problems show that the method works very competitively against other well known modifications of Newton's method such as Adaptive Cubic Regularization and BFGS, as well as first order methods such as Unbounded Two-way Backtracking Gradient Descent. In this paper, we resolve the convergence guarantee issue by proposing a modification of New Q-Newton's method, named New Q-Newton's method Backtracking, which incorporates a more sophisticated use of hyperparameters and a Backtracking line search. This new method has very good theoretical guarantees, which for a {\bf Morse function} yields the following (which is unknown for New Q-Newton's method): {\bf Theorem.} Let $f:\mathbb{R}^m\rightarrow \mathbb{R}$ be a Morse function, that is all its critical points have invertible Hessian. Then for a sequence $\{x_n\}$ constructed by New Q-Newton's method Backtracking from a random initial point $x_0$, we have the following two alternatives: i) $\lim _{n\rightarrow\infty}||x_n||=\infty$, or ii) $\{x_n\}$ converges to a point $x_{\infty}$ which is a {\bf local minimum} of $f$, and the rate of convergence is {\bf quadratic}. Moreover, if $f$ has compact sublevels, then only case ii) happens. As far as we know, for Morse functions, this is the best theoretical guarantee for iterative optimization algorithms so far in the literature. We have tested in experiments on small scale, with some further simplified versions of New Q-Newton's method Backtracking, and found that the new method significantly improve New Q-Newton's method.

In cooperative robot teams, communications can speed up execution, reduce duplication, and prevent interference. Although many systems use explicit communications, persistent intra-team digital communications is not guaranteed. One approach to this challenge is to use implicit communication to infer state rather than using digital messages. We investigate using implicit communication in the form of observation to infer state to coordinate a robot team in a coverage task. We demonstrate how pruning and backtracking a search tree can improve multi-robot coverage. Experiments were conducted to compare team performance of a robot team using observation-based cooperation to one that uses explicit communications.