Coping with software defects that occur in the post-deployment stage is a challenging problem: bugs may occur only when the system uses a specific configuration and only under certain usage scenarios. Nevertheless, halting production systems until the bug is tracked and fixed is often impossible. Thus, developers have to try to reproduce the bug in laboratory conditions. Often the reproduction of the bug consists of the lion share of the debugging effort.
Load balancing is a technique which allows efficient parallelization of irregular workloads, and a key component of many applications and parallelizing runtimes. Work-stealing is a popular technique for implementing load balancing, where each parallel thread maintains its own work set of items and occasionally steals items from the sets of other threads.
Recently, powerful Large Language Models (LLMs) have become easily accessible to hundreds of millions of users worldwide. However, their strong capabilities and vast world knowledge do not come without associated privacy risks. In this work, we focus on the emerging privacy threat LLMs pose - the ability to accurately infer personal information from online texts. Despite the growing importance of LLM-based author profiling, research in this area has been hampered by a lack of suitable public datasets, largely due to ethical and privacy concerns associated with real personal data. In this work, we take two steps to address this problem: (i) we construct a simulation framework for the popular social media platform Reddit using LLM agents seeded with synthetic personal profiles; (ii) using this framework, we generate SynthPAI, a diverse synthetic dataset of over 7800 comments manually labeled for personal attributes. We validate our dataset with a human study showing that humans barely outperform random guessing on the task of distinguishing our synthetic comments from real ones. Further, we verify that our dataset enables meaningful personal attribute inference research by showing across 18 state-of-the-art LLMs that our synthetic comments allow us to draw the same conclusions as real-world data. Together, this indicates that our dataset and pipeline provide a strong and privacy-preserving basis for future research toward understanding and mitigating the inference-based privacy threats LLMs pose.
Building correct and efficient concurrent algorithms is known to be a difficult problem of fundamental importance. To achieve efficiency, designers try to remove unnecessary and costly synchronization. However, not only is this manual trial-and-error process ad-hoc, time consuming and error-prone, but it often leaves designers pondering the question of: is it inherently impossible to eliminate certain synchronization, or is it that I was unable to eliminate it on this attempt and I should keep trying? In this paper we respond to this question. We prove that it is impossible to build concurrent implementations of classic and ubiquitous specifications such as sets, queues, stacks, mutual exclusion and read-modify-write operations, that completely eliminate the use of expensive synchronization. We prove that one cannot avoid the use of either: i) read-after-write (RAW), where a write to shared variable A is followed by a read to a different shared variable B without a write to B in between, or ii) atomic write-after-read (AWAR), where an atomic operation reads and then writes to shared locations. Unfortunately, enforcing RAW or AWAR is expensive on all current mainstream processors. To enforce RAW, memory ordering--also called fence or barrier--instructions must be used. To enforce AWAR, atomic instructions such as compare-and-swap are required. However, these instructions are typically substantially slower than regular instructions. Although algorithm designers frequently struggle to avoid RAW and AWAR, their attempts are often futile. Our result characterizes the cases where avoiding RAW and AWAR is impossible. On the flip side, our result can be used to guide designers towards new algorithms where RAW and AWAR can be eliminated.
Randomized Smoothing (RS) is considered the state-of-the-art approach to obtain certifiably robust models for challenging tasks. However, current RS approaches drastically decrease standard accuracy on unperturbed data, severely limiting their real-world utility. To address this limitation, we propose a compositional architecture, ACES, which certifiably decides on a per-sample basis whether to use a smoothed model yielding predictions with guarantees or a more accurate standard model without guarantees. This, in contrast to prior approaches, enables both high standard accuracies and significant provable robustness. On challenging tasks such as ImageNet, we obtain, e.g., $80.0\%$ natural accuracy and $28.2\%$ certifiable accuracy against $\ell_2$ perturbations with $r=1.0$. We release our code and models at https://github.com/eth-sri/aces.
We present a novel approach for synthesizing robust relational layouts from examples. Given an application design consisting of a set of views and their location on the screen, we synthesize a relational layout that when rendered, places the components at that same location. We present an end-to-end system, called InferUI, that addresses the above challenge in the context of Android. The system is based on the following technical contributions: (i) a formalization of the latest and most efficient ConstraintLayout class, capturing a rich set of relational constraints, (ii) a set of robustness properties designed to prevent common layout generalization errors, (iii) a synthesis algorithm that produces relational layouts that generalize across multiple screen sizes and resolutions, and (iv) a probabilistic model of constraints that guides the synthesizer towards layouts preferred by developers. Our evaluation shows that InferUI is practically effective: it successfully synthesizes real world complex layouts obtained from top 500 GitHub and top 500 Google Play Store applications, succeeds in 100% of the cases when synthesizing layouts for a single device, and correctly generalizes 92% of the views across multiple devices, all without requiring additional specifications.
Large language models have demonstrated outstanding performance on a wide range of tasks such as question answering and code generation. On a high level, given an input, a language model can be used to automatically complete the sequence in a statistically-likely way. Based on this, users prompt these models with language instructions or examples, to implement a variety of downstream tasks. Advanced prompting methods can even imply interaction between the language model, a user, and external tools such as calculators. However, to obtain state-of-the-art performance or adapt language models for specific tasks, complex task- and model-specific programs have to be implemented, which may still require ad-hoc interaction. Based on this, we present the novel idea of Language Model Programming (LMP). LMP generalizes language model prompting from pure text prompts to an intuitive combination of text prompting and scripting. Additionally, LMP allows constraints to be specified over the language model output. This enables easy adaption to many tasks while abstracting language model internals and providing high-level semantics. To enable LMP, we implement LMQL(short for Language Model Query Language), which leverages the constraints and control flow from an LMP prompt to generate an efficient inference procedure that minimizes the number of expensive calls to the underlying language model. We show that LMQL can capture a wide range of state-of-the-art prompting methods in an intuitive way, especially facilitating interactive flows that are challenging to implement with existing high-level APIs. Our evaluation shows that we retain or increase the accuracy on several downstream tasks, while also significantly reducing the required amount of computation or cost in the case of pay-to-use APIs (26-85% cost savings).
Refactoring has become an integral part of modern software development, with wide support in popular integrated development environments (IDEs). Modern IDEs provide a fixed set of supported refactorings, listed in a refactoring menu. But with IDEs supporting more and more refactorings, it is becoming increasingly difficult for programmers to discover and memorize all their names and meanings. Also, since the set of refactorings is hard-coded, if a programmer wants to achieve a slightly different code transformation, she has to either apply a (possibly non-obvious) sequence of several built-in refactorings, or just perform the transformation by hand. We propose a novel approach to refactoring, based on synthesis from examples, which addresses these limitations. With our system, the programmer need not worry how to invoke individual refactorings or the order in which to apply them. Instead, a transformation is achieved via three simple steps: the programmer first indicates the start of a code refactoring phase; then she performs some of the desired code changes manually; and finally, she asks the tool to complete the refactoring. Our system completes the refactoring by first extracting the difference between the starting program and the modified version, and then synthesizing a sequence of refactorings that achieves (at least) the desired changes. To enable scalable synthesis, we introduce local refactorings , which allow for first discovering a refactoring sequence on small program fragments and then extrapolating it to a full refactoring sequence. We implemented our approach as an Eclipse plug-in, with an architecture that is easily extendable with new refactorings. The experimental results are encouraging: with only minimal user input, the synthesizer was able to quickly discover complex refactoring sequences for several challenging realistic examples.
The widespread applicability of large language models (LLMs) has increased the availability of many fine-tuned models of various sizes targeting specific tasks. Given a set of such specialized models, to maximize overall performance, it is important to figure out the optimal strategy for selecting the right model for a given user query. An effective strategy could drastically increase overall performance and even offer improvements over a single large monolithic model. Existing approaches typically fall into two categories: routing, where a single model is selected for each query, and cascading, which runs a sequence of increasingly larger models until a satisfactory answer is obtained. However, both have notable limitations: routing commits to an initial model without flexibility, while cascading requires executing every model in sequence, which can be inefficient. Additionally, the conditions under which these strategies are provably optimal remain unclear. In this work, we derive optimal strategies for both routing and cascading. Building on this analysis, we propose a novel approach called cascade routing, which combines the adaptability of routing with the cost-efficiency of cascading. Our experiments demonstrate that cascade routing consistently outperforms both routing and cascading across a variety of settings, improving both output quality and lowering computational cost, thus offering a unified and efficient solution to the model selection problem.
