arXiv subjects

Eitan Farchi

Publications and source records attributed to Eitan Farchi.

At least 37 records · Page 2Linked to original sources

arXiv2024

Generating Unseen Code Tests In Infinitum

Large Language Models (LLMs) are used for many tasks, including those related to coding. An important aspect of being able to utilize LLMs is the ability to assess their fitness for specific usages. The common practice is to evaluate LLMs against a set of benchmarks. While benchmarks provide a sound foundation for evaluation and comparison of alternatives, they suffer from the well-known weakness of leaking into the training data \cite{Xu2024Benchmarking}. We present a method for creating benchmark variations that generalize across coding tasks and programming languages, and may also be applied to in-house code bases. Our approach enables ongoing generation of test-data thus mitigating the leaking into the training data issue. We implement one benchmark, called \textit{auto-regression}, for the task of text-to-code generation in Python. Auto-regression is specifically created to aid in debugging and in tracking model generation changes as part of the LLM regression testing process.

cs.AI↗

arXiv2024

Effective Technical Reviews

There are two ways to check if a program is correct, namely execute it or review it. While executing a program is the ultimate test for its correctness reviewing the program can occur earlier in its development and find problems if done effectively. This work focuses on review techniques. It enables the programmer to effectively review a program and find a range of problems from concurrency to interface issues. The review techniques can be applied in a time constrained industrial development context and are enhanced by knowledge on programming pitfalls.

cs.SE↗

arXiv2024

Using Combinatorial Optimization to Design a High quality LLM Solution

We introduce a novel LLM based solution design approach that utilizes combinatorial optimization and sampling. Specifically, a set of factors that influence the quality of the solution are identified. They typically include factors that represent prompt types, LLM inputs alternatives, and parameters governing the generation and design alternatives. Identifying the factors that govern the LLM solution quality enables the infusion of subject matter expert knowledge. Next, a set of interactions between the factors are defined and combinatorial optimization is used to create a small subset $P$ that ensures all desired interactions occur in $P$. Each element $p \in P$ is then developed into an appropriate benchmark. Applying the alternative solutions on each combination, $p \in P$ and evaluating the results facilitate the design of a high quality LLM solution pipeline. The approach is especially applicable when the design and evaluation of each benchmark in $P$ is time-consuming and involves manual steps and human evaluation. Given its efficiency the approach can also be used as a baseline to compare and validate an autoML approach that searches over the factors governing the solution.

cs.CL↗

arXiv2024

Alignment Studio: Aligning Large Language Models to Particular Contextual Regulations

The alignment of large language models is usually done by model providers to add or control behaviors that are common or universally understood across use cases and contexts. In contrast, in this article, we present an approach and architecture that empowers application developers to tune a model to their particular values, social norms, laws and other regulations, and orchestrate between potentially conflicting requirements in context. We lay out three main components of such an Alignment Studio architecture: Framers, Instructors, and Auditors that work in concert to control the behavior of a language model. We illustrate this approach with a running example of aligning a company's internal-facing enterprise chatbot to its business conduct guidelines.

cs.CL↗

arXiv2024

Quality Engineering for Agile and DevOps on the Cloud and Edge

Today's software projects include enhancements, fixes, and patches need to be delivered almost on a daily basis to clients. Weekly and daily releases are pretty much the norm and sit alongside larger feature upgrades and quarterly releases. Software delivery has to be more agile now than ever before. Companies that were, in the past, experimenting with agile based delivery models, are now looking to scale it to enterprise grade. This shifts the need from the ability to build and execute tests rapidly, to using different means, technologies and procedures to provide rapid and insightful validation sequences and tests to establish quality withing the manufacturing cycle. This book addresses the need of effectively embedding quality engineering throughout the agile development cycle thus addressing the need for enterprise scale high quality agile development.

cs.SE↗

arXiv2023

Unveiling Safety Vulnerabilities of Large Language Models

As large language models become more prevalent, their possible harmful or inappropriate responses are a cause for concern. This paper introduces a unique dataset containing adversarial examples in the form of questions, which we call AttaQ, designed to provoke such harmful or inappropriate responses. We assess the efficacy of our dataset by analyzing the vulnerabilities of various models when subjected to it. Additionally, we introduce a novel automatic approach for identifying and naming vulnerable semantic regions - input semantic areas for which the model is likely to produce harmful outputs. This is achieved through the application of specialized clustering techniques that consider both the semantic similarity of the input attacks and the harmfulness of the model's responses. Automatically identifying vulnerable semantic regions enhances the evaluation of model weaknesses, facilitating targeted improvements to its safety mechanisms and overall reliability.

cs.CL↗

arXiv2023

Predicting Question-Answering Performance of Large Language Models through Semantic Consistency

Semantic consistency of a language model is broadly defined as the model's ability to produce semantically-equivalent outputs, given semantically-equivalent inputs. We address the task of assessing question-answering (QA) semantic consistency of contemporary large language models (LLMs) by manually creating a benchmark dataset with high-quality paraphrases for factual questions, and release the dataset to the community. We further combine the semantic consistency metric with additional measurements suggested in prior work as correlating with LLM QA accuracy, for building and evaluating a framework for factual QA reference-less performance prediction -- predicting the likelihood of a language model to accurately answer a question. Evaluating the framework on five contemporary LLMs, we demonstrate encouraging, significantly outperforming baselines, results.

cs.CL↗

arXiv2023

Generalized Coverage Criteria for Combinatorial Sequence Testing

We present a new model-based approach for testing systems that use sequences of actions and assertions as test vectors. Our solution includes a method for quantifying testing quality, a tool for generating high-quality test suites based on the coverage criteria we propose, and a framework for assessing risks. For testing quality, we propose a method that specifies generalized coverage criteria over sequences of actions, which extends previous approaches. Our publicly available tool demonstrates how to extract effective test suites from test plans based on these criteria. We also present a Bayesian approach for measuring the probabilities of bugs or risks, and show how this quantification can help achieve an informed balance between exploitation and exploration in testing. Finally, we provide an empirical evaluation demonstrating the effectiveness of our tool in finding bugs, assessing risks, and achieving coverage.

cs.SE↗

arXiv2023

Characterizing how 'distributional' NLP corpora distance metrics are

A corpus of vector-embedded text documents has some empirical distribution. Given two corpora, we want to calculate a single metric of distance (e.g., Mauve, Frechet Inception) between them. We describe an abstract quality, called `distributionality', of such metrics. A non-distributional metric tends to use very local measurements, or uses global measurements in a way that does not fully reflect the distributions' true distance. For example, if individual pairwise nearest-neighbor distances are low, it may judge the two corpora to have low distance, even if their two distributions are in fact far from each other. A more distributional metric will, in contrast, better capture the distributions' overall distance. We quantify this quality by constructing a Known-Similarity Corpora set from two paraphrase corpora and calculating the distance between paired corpora from it. The distances' trend shape as set element separation increases should quantify the distributionality of the metric. We propose that Average Hausdorff Distance and energy distance between corpora are representative examples of non-distributional and distributional distance metrics, to which other metrics can be compared, to evaluate how distributional they are.

cs.CL↗

arXiv2023

Data Drift Monitoring for Log Anomaly Detection Pipelines

Logs enable the monitoring of infrastructure status and the performance of associated applications. Logs are also invaluable for diagnosing the root causes of any problems that may arise. Log Anomaly Detection (LAD) pipelines automate the detection of anomalies in logs, providing assistance to site reliability engineers (SREs) in system diagnosis. Log patterns change over time, necessitating updates to the LAD model defining the `normal' log activity profile. In this paper, we introduce a Bayes Factor-based drift detection method that identifies when intervention, retraining, and updating of the LAD model are required with human involvement. We illustrate our method using sequences of log activity, both from unaltered data, and simulated activity with controlled levels of anomaly contamination, based on real collected log data.

cs.LG↗

arXiv2023

Automatic Generation of Attention Rules For Containment of Machine Learning Model Errors

Machine learning (ML) solutions are prevalent in many applications. However, many challenges exist in making these solutions business-grade. For instance, maintaining the error rate of the underlying ML models at an acceptably low level. Typically, the true relationship between feature inputs and the target feature to be predicted is uncertain, and hence statistical in nature. The approach we propose is to separate the observations that are the most likely to be predicted incorrectly into 'attention sets'. These can directly aid model diagnosis and improvement, and be used to decide on alternative courses of action for these problematic observations. We present several algorithms (`strategies') for determining optimal rules to separate these observations. In particular, we prefer strategies that use feature-based slicing because they are human-interpretable, model-agnostic, and require minimal supplementary inputs or knowledge. In addition, we show that these strategies outperform several common baselines, such as selecting observations with prediction confidence below a threshold. To evaluate strategies, we introduce metrics to measure various desired qualities, such as their performance, stability, and generalizability to unseen data; the strategies are evaluated on several publicly-available datasets. We use TOPSIS, a Multiple Criteria Decision Making method, to aggregate these metrics into a single quality score for each strategy, to allow comparison.

cs.LG↗

arXiv2023

Convex Bounds on the Softmax Function with Applications to Robustness Verification

The softmax function is a ubiquitous component at the output of neural networks and increasingly in intermediate layers as well. This paper provides convex lower bounds and concave upper bounds on the softmax function, which are compatible with convex optimization formulations for characterizing neural networks and other ML models. We derive bounds using both a natural exponential-reciprocal decomposition of the softmax as well as an alternative decomposition in terms of the log-sum-exp function. The new bounds are provably and/or numerically tighter than linear bounds obtained in previous work on robustness verification of transformers. As illustrations of the utility of the bounds, we apply them to verification of transformers as well as of the robustness of predictive uncertainty estimates of deep ensembles.

cs.LG↗

arXiv2022

Measuring the Measuring Tools: An Automatic Evaluation of Semantic Metrics for Text Corpora

The ability to compare the semantic similarity between text corpora is important in a variety of natural language processing applications. However, standard methods for evaluating these metrics have yet to be established. We propose a set of automatic and interpretable measures for assessing the characteristics of corpus-level semantic similarity metrics, allowing sensible comparison of their behavior. We demonstrate the effectiveness of our evaluation measures in capturing fundamental characteristics by evaluating them on a collection of classical and state-of-the-art metrics. Our measures revealed that recently-developed metrics are becoming better in identifying semantic distributional mismatch while classical metrics are more sensitive to perturbations in the surface text levels.

cs.CL↗

arXiv2022

Random Test Generation of Application Programming Interfaces

Cloud high quality API (Application Programming Interface) testing is essential for supporting the API economy. Autotest is a random test generator that addresses this need. It reads the API specification and deduces a model used in the test generation. This paper describes Autotest. It also address the topic of API specification pitfalls which Autotest may reveal when reading the specification. A best practice is to add an appropriate test to the regression once a problem is revealed and solved. How to do that in the context of Autotest's random test generation is covered.

cs.SE↗

arXiv2022

Detection of data drift and outliers affecting machine learning model performance over time

A trained ML model is deployed on another `test' dataset where target feature values (labels) are unknown. Drift is distribution change between the training and deployment data, which is concerning if model performance changes. For a cat/dog image classifier, for instance, drift during deployment could be rabbit images (new class) or cat/dog images with changed characteristics (change in distribution). We wish to detect these changes but can't measure accuracy without deployment data labels. We instead detect drift indirectly by nonparametrically testing the distribution of model prediction confidence for changes. This generalizes our method and sidesteps domain-specific feature representation. We address important statistical issues, particularly Type-1 error control in sequential testing, using Change Point Models (CPMs; see Adams and Ross 2012). We also use nonparametric outlier methods to show the user suspicious observations for model diagnosis, since the before/after change confidence distributions overlap significantly. In experiments to demonstrate robustness, we train on a subset of MNIST digit classes, then insert drift (e.g., unseen digit class) in deployment data in various settings (gradual/sudden changes in the drift proportion). A novel loss function is introduced to compare the performance (detection delay, Type-1 and 2 errors) of a drift detector under different levels of drift class contamination.

stat.AP↗

arXiv2022

Using Fuzzy Matching of Queries to optimize Database workloads

Directed Acyclic Graphs (DAGs) are commonly used in Databases and Big Data computational engines like Apache Spark for representing the execution plan of queries. We refer to such graphs as Query Directed Acyclic Graphs (QDAGs). This paper uses similarity hashing to arrive at a fingerprint such that the fingerprint embodies the compute requirements of the query for QDAGs. The fingerprint, thus obtained, can be used to predict the runtime behaviour of a query based on queries executed in the past having similar QDAGs. We discuss two approaches to arrive at a fingerprint, their pros and cons and how aspects of both approaches can be combined to improve the predictions. Using a hybrid approach, we demonstrate that we are able to predict runtime behaviour of a QDAG with more than 80% accuracy.

cs.DB↗

arXiv2022

High-quality Conversational Systems

Conversational systems or chatbots are an example of AI-Infused Applications (AIIA). Chatbots are especially important as they are often the first interaction of clients with a business and are the entry point of a business into the AI (Artificial Intelligence) world. The quality of the chatbot is, therefore, key. However, as is the case in general with AIIAs, it is especially challenging to assess and control the quality of chatbot systems. Beyond the inherent statistical nature of these systems, where occasional failure is acceptable, we identify two major challenges. The first is to release an initial system that is of sufficient quality such that humans will interact with it. The second is to maintain the quality, enhance its capabilities, improve it and make necessary adjustments based on changing user requests or drift. These challenges exist because it is impossible to predict the real distribution of user requests and the natural language they will use to express these requests. Moreover, any empirical distribution of requests is likely to change over time. This may be due to periodicity, changing usage, and drift of topics. We provide a methodology and set of technologies to address these challenges and to provide automated assistance through a human-in-the-loop approach. We notice that it is crucial to connect between the different phases in the lifecycle of the chatbot development and to make sure it provides its expected business value. For example, that it frees human agents to deal with tasks other than answering human users. Our methodology and technologies apply during chatbot training in the pre-production phase, through to chatbot usage in the field in the post-production phase. They implement the `test first' paradigm by assisting in agile design, and support continuous integration through actionable insights.

cs.HC↗

arXiv2022

Theory and Practice of Quality Assurance for Machine Learning Systems An Experiment Driven Approach

The crafting of machine learning (ML) based systems requires statistical control throughout its life cycle. Careful quantification of business requirements and identification of key factors that impact the business requirements reduces the risk of a project failure. The quantification of business requirements results in the definition of random variables representing the system key performance indicators that need to be analyzed through statistical experiments. In addition, available data for training and experiment results impact the design of the system. Once the system is developed, it is tested and continually monitored to ensure it meets its business requirements. This is done through the continued application of statistical experiments to analyze and control the key performance indicators. This book teaches the art of crafting and developing ML based systems. It advocates an "experiment first" approach stressing the need to define statistical experiments from the beginning of the project life cycle. It also discusses in detail how to apply statistical control on the ML based system throughout its lifecycle.

cs.LG↗