Vision Language Models (VLMs) are foundational for low-latency, privacy-preserving on-device AI in real-time applications like UI agents and VQA. The VLM prefilling phase, which processes the entire visual-textual input, faces the critical challenge of a long Time-to-First-Token (TTFT). One promising approach to reduce TTFT is to exploit the temporal locality by reusing block-level computations across consecutive frames. Unfortunately, current Transformer-based VLMs break the spatial invariance of CNNs and invalidate the strict-prefix KV-cache mechanism of decoder-only LLMs; in practice, even a single-pixel mismatch can prevent reuse.

Docker containers are being widely used in large-scale industrial environments. In practice, developers must manually specify the base image in the dockerfile in the process of container creation. However, finding the proper base image is a nontrivial task because manually searching is time-consuming and easily leads to the use of unsuitable base images, especially for newcomers. There is still a lack of automatic approaches for recommending related base image for developers through dockerfile configuration. To tackle this problem, this paper makes the first attempt to propose a neural network approach named DCCimagerec which is based on deep configuration comprehension. It aims to use the structural configuration features of dockerfile extracted by AST and path-attention model to recommend potentially suitable base image. The evaluation experiments based on about 83,000 dockerfiles show that DCCimagerec outperforms multiple baselines, improving Precision by 7.5%-67.5%, Recall by 6.2%-106.6%, and F1 by 7.5%-150.2%.

Over the past several years, the Robot Operating System (ROS), has grown from a small research project into the most popular framework for robotics development.It offers a core set of software for operating robots that can be extended by creating or using existing packages, making it possible to program robotic software that can be reused on different hardware platforms.With thousands of packages available per stable distribution, encapsulating algorithms, sensor drivers, etc., it is the de facto middleware for robotics.However, finding the proper ROS package is a nontrivial task because ROS packages involve different functions and even with the same function, there are different ROS packages for different tasks.So it is timeconsuming for developers to find suitable ROS packages for given task, especially for newcomers.To tackle this challenge, we build a ROS package knowledge graph, ROSKG, including the basic information of ROS packages and ROS package characteristics extracted from text descriptions, to comprehensively and precisely characterize ROS packages.Based on ROSKG, we support ROS packages search with specific task description or attributes as input.A comprehensive evaluation of ROSKG shows the high accuracy of our knowledge construction approach.A user study shows that ROSKG is promising in helping developers find suitable ROS packages for robotics software development tasks.

Container technologies are being widely used in large scale production cloud environments, of which Docker has become the de-facto industry standard. As a key step, containers need to define their dependent base image, which makes complex dependencies exist in a large number of containers. Prior studies have shown that references between software packages could form technical dependencies, thus forming a dependency network. However, little is known about the details of docker container dependency networks. In this paper, we perform an empirical study on the dependency network of docker containers from more than 120,000 dockerfiles. We construct the container dependency network and analyze its network structure. Further, we focus on the Top-100 dominant containers and investigate their subnetworks, including diversity and relationships. Our findings help to characterize and understand the container dependencies in the docker community and motivate the need for developing container dependency management tools.