Danmaku, users' live comments synchronized with, and overlaying on videos, has recently shown potential in promoting online video-based learning.However, user-generated danmaku can be scarce-especially in newer or less viewed videos-and its quality is unpredictable, limiting its educational impact.This paper explores how large multimodal models (LMM) can be leveraged to automatically generate effective, high-quality danmaku.We first conducted a formative study to identify the desirable characteristics of contentand emotion-related danmaku in educational videos.Based on the obtained insights, we developed ClassComet, an educational video platform with novel LMM-driven techniques for generating relevant types of danmaku to enhance video-based learning.Through user studies, we examined the quality of generated danmaku and their influence on learning experiences.The results indicate that our generated danmaku is comparable to human-created ones, and videos with both content-and emotion-related danmaku showed significant improvement in viewers' engagement and learning outcome.

LLM-to-NAS is a promising field at the intersection of Large Language Models (LLMs) and Neural Architecture Search (NAS), as recent research has explored the potential of architecture generation leveraging LLMs on multiple search spaces. However, the existing LLM-to-NAS methods face the challenges of limited search spaces, time-cost search efficiency, and uncompetitive performance across standard NAS benchmarks and multiple downstream tasks. In this work, we propose the Reflective Zero-cost NAS (RZ-NAS) method that can search NAS architectures with humanoid reflections and training-free metrics to elicit the power of LLMs. We rethink LLMs’ roles in NAS in current work and design a structured, prompt-based to comprehensively understand the search tasks and architectures from both text and code levels. By integrating LLM reflection modules, we use LLM-generated feedback to provide linguistic guidance within architecture optimization. RZ-NAS enables effective search within both micro and macro search spaces without extensive time cost, achieving SOTA performance across multiple downstream tasks.

Differentiable NAS (DARTS) is a simple and efficient neural architecture search method that has been extensively adopted in various machine learning tasks. % Nevertheless, DARTS still encounters several robustness issues, mainly the domination of skip connections. % The resulting architectures are full of parametric-free operations, leading to performance collapse. % Existing methods suggest that the skip connection has additional advantages in optimization compared to other parametric operations and propose to alleviate the domination of skip connections by eliminating these additional advantages. % In this paper, we analyze this issue from a simple and straightforward perspective and propose that the domination of skip connections results from parametric operations overfitting the training data while architecture parameters are trained on the validation data, leading to undesired behaviors. % Based on this observation, we propose the operation-level early stopping (OLES) method to overcome this issue and robustify DARTS without introducing any computation overhead. % Extensive experimental results can verify our hypothesis and the effectiveness of OLES.