The field of natural language processing is moving towards more effective and efficient alignment of large language models (LLMs) with human preferences. Recent developments have focused on improving the quality and diversity of training data, as well as developing new methods for optimizing LLMs. One key area of research is the use of multi-objective alignment, which allows LLMs to balance multiple competing objectives, such as informativeness and conciseness. Another important area is the development of more effective and efficient methods for generating and selecting high-quality training data, including the use of synthetic data and data selection principles. Additionally, researchers are exploring new approaches to prompt engineering, including the use of controlled natural language and inference-aware prompt optimization. Noteworthy papers in this area include: Pareto Multi-Objective Alignment for Language Models, which proposes a principled and computationally efficient algorithm for multi-objective alignment. Beyond Single: A Data Selection Principle for LLM Alignment via Fine-Grained Preference Signals, which introduces a novel data selection principle that advocates for selecting a subset of high-consensus data for efficient training. Inference-Aware Prompt Optimization for Aligning Black-Box Large Language Models, which introduces a unified framework that jointly optimizes the prompt and inference scale, while being aware of the inference budget and different task objectives.