Search

The metadata involved in integrating code, data, configuration, and feedback into predictive models is varied and complex. This complexity is further compounded by the agile development practices favored by data scientists and machine learning engineers. These practices emphasize high experimentation velocity and frequent deployments, which can make it challenging to keep track of all the relevant metadata. The iterative nature of agile methods means that models, datasets, and configurations are constantly evolving. Each experiment might involve tweaks to the data preprocessing steps, changes in model hyperparameters, or updates to the deployment environment. The need for rapid iteration can lead to shortcuts or oversights in documentation and metadata management. Effective metadata management requires robust yet flexible tools and practices that can integrate and organize this information without slowing down the development process. Traditional context management often emphasizes a ``metadata first'' approach, which can introduce significant friction for developers. FlorDB reduces this friction through multiversion hindsight logging and incremental context maintenance, allowing developers to add and refine metadata after the fact. This ``metadata later'' approach enables a more flexible and incremental development process, allowing data scientists to focus on model creation and refinement without the burden of documentation upfront. As shown in a demo, FlorDB can be used to build AI/ML applications with integrated train-infer pipelines and managed feedback loops. Ultimately, the goal of FlorDB is to ensure that critical metadata is maintained accurately and efficiently, even in fast-paced agile workflows.

In this work, firstly, we propose how to experimentally demonstrate the single photon steering phenomenon in a simple way. The quantum steering phenomenon was discovered by Erwin Schrodinger, who reason that the laws of quantum mechanics obliges us to admit that by suitable measurements taken on one of two entangled system, then the state of the other system can be determined and steered. On the other hand, the first proposal about the nonlocal property of a single photon focus on showing the Bell nonlocality by using the single-photon path entanglement. Here, we propose a new experimental scheme that, by incorporating and manipulating the internal degree of freedom (IDF) of the photon, easily demonstrate the nonlocal steering phenomenon of single-photon's states. The experimental set-up that we propose differs from those published in the quantum optics' literature to shown the single photon nonlocality, which generally use the path entanglement of photons and homodyne measurements. Secondly, by incorporating its IDF, we show that the single photon steering and the quantum mechanical interaction-free measurement (QM-IFM) are identical phenomena; we will argue that QM-IFM is just a particular instance of the single-photon steering. In short, both the single photon steering and QM-IFM (being identical phenomena) have a common cause: the nonlocality of the wave function. In conclusion, we have demonstrated that it is possible to produce a nonlocal phenomena by manipulating the single photon internal degree of freedom. On the other hand, this unification between steering and QM-IFM could establish a strong support to counterfactual quantum communication and computation., Comment: 26 pages, 7 figures. submitted. arXiv admin note: text overlap with arXiv:2109.01058

Interactive notebooks, such as Jupyter, have revolutionized the field of data science by providing an integrated environment for data, code, and documentation. However, their adoption by robotics researchers and model developers has been limited. This study investigates the logging and record-keeping practices of robotics researchers, drawing parallels to the pre-interactive notebook era of data science. Through interviews with robotics researchers, we identified the reliance on diverse and often incompatible tools for managing experimental data, leading to challenges in reproducibility and data traceability. Our findings reveal that robotics researchers can benefit from a specialized version of interactive notebooks that supports comprehensive data entry, continuous context capture, and agile data staging. We propose extending interactive notebooks to better serve the needs of robotics researchers by integrating features akin to traditional lab notebooks. This adaptation aims to enhance the organization, analysis, and reproducibility of experimental data in robotics, fostering a more streamlined and efficient research workflow.

Organizations rely on machine learning engineers (MLEs) to deploy models and maintain ML pipelines in production. Due to models' extensive reliance on fresh data, the operationalization of machine learning, or MLOps, requires MLEs to have proficiency in data science and engineering. When considered holistically, the job seems staggering -- how do MLEs do MLOps, and what are their unaddressed challenges? To address these questions, we conducted semi-structured ethnographic interviews with 18 MLEs working on various applications, including chatbots, autonomous vehicles, and finance. We find that MLEs engage in a workflow of (i) data preparation, (ii) experimentation, (iii) evaluation throughout a multi-staged deployment, and (iv) continual monitoring and response. Throughout this workflow, MLEs collaborate extensively with data scientists, product stakeholders, and one another, supplementing routine verbal exchanges with communication tools ranging from Slack to organization-wide ticketing and reporting systems. We introduce the 3Vs of MLOps: velocity, visibility, and versioning -- three virtues of successful ML deployments that MLEs learn to balance and grow as they mature. Finally, we discuss design implications and opportunities for future work., Comment: arXiv admin note: text overlap with arXiv:2209.09125

Production Machine Learning involves continuous training: hosting multiple versions of models over time, often with many model versions running at once. When model performance does not meet expectations, Machine Learning Engineers (MLEs) debug issues by exploring and analyzing numerous prior versions of code and training data to identify root causes and mitigate problems. Traditional debugging and logging tools often fall short in managing this experimental, multi-version context. FlorDB introduces Multiversion Hindsight Logging, which allows engineers to use the most recent version's logging statements to query past versions, even when older versions logged different data. Log statement propagation enables consistent injection of logging statements into past code versions, regardless of changes to the codebase. Once log statements are propagated across code versions, the remaining challenge in Multiversion Hindsight Logging is to efficiently replay the new log statements based on checkpoints from previous runs. Finally, a coherent user experience is required to help MLEs debug across all versions of code and data. To this end, FlorDB presents a unified relational model for efficient handling of historical queries, offering a comprehensive view of the log history to simplify the exploration of past code iterations. We present a performance evaluation on diverse benchmarks confirming its scalability and the ability to deliver real-time query responses, leveraging query-based filtering and checkpoint-based parallelism for efficient replay.

Organizations rely on machine learning engineers (MLEs) to operationalize ML, i.e., deploy and maintain ML pipelines in production. The process of operationalizing ML, or MLOps, consists of a continual loop of (i) data collection and labeling, (ii) experimentation to improve ML performance, (iii) evaluation throughout a multi-staged deployment process, and (iv) monitoring of performance drops in production. When considered together, these responsibilities seem staggering -- how does anyone do MLOps, what are the unaddressed challenges, and what are the implications for tool builders? We conducted semi-structured ethnographic interviews with 18 MLEs working across many applications, including chatbots, autonomous vehicles, and finance. Our interviews expose three variables that govern success for a production ML deployment: Velocity, Validation, and Versioning. We summarize common practices for successful ML experimentation, deployment, and sustaining production performance. Finally, we discuss interviewees' pain points and anti-patterns, with implications for tool design., Comment: 20 pages, 4 figures

Coordination protocols help programmers of distributed systems reason about the effects of transactions on the state of the system, but they're not cheap. Coordination protocols may involve multiple rounds of communication, which can hurt system responsiveness. There exist many efforts in distributed computing for managing the coordination-performance trade-off. More recent is a line of work that characterizes the class of workloads for which coordination is not necessary for consistency: namely, logically monotonic programs. In this paper, we present a case study of logical monotonicity in workloads typical to computational biology. We leverage the Bloom language to write efficient distributed programs, and compare their performance to equivalent programs written in UPC++, a popular language for writing distributed programs. Additionally, we leverage Bloom's analysis tools to identify points-of-coordination, and use our own experience using Bloom to recommend some higher-level abstractions for users without strong distributed computing backgrounds.

In modern Machine Learning, model training is an iterative, experimental process that can consume enormous computation resources and developer time. To aid in that process, experienced model developers log and visualize program variables during training runs. Exhaustive logging of all variables is infeasible. Optimistic logging can be accompanied by program checkpoints; this allows developers to add log statements post-hoc, and "replay" desired log statements from checkpoint -- a process we refer to as hindsight logging. Unfortunately, hindsight logging raises tricky problems in data management and software engineering. Done poorly, hindsight logging can waste resources and generate technical debt embodied in multiple variants of training code. In this paper, we present methodologies for efficient and effective logging practices for model training, with a focus on techniques for hindsight logging. Our goal is for experienced model developers to learn and adopt these practices. To make this easier, we provide an open-source suite of tools for Fast Low-Overhead Recovery (flor) that embodies our design across three tasks: (i) efficient background logging in Python, (ii) adaptable periodic checkpointing, and (iii) an instrumentation library that codifies hindsight logging for efficient and automatic record-replay of model-training. Model developers can use each flor tool separately as they see fit, or they can use flor in hands-free mode, entrusting it to instrument their code end-to-end for efficient record-replay. Our solutions leverage techniques from physiological transaction logs and recovery in database systems. Evaluations on modern ML benchmarks demonstrate that flor can produce fast checkpointing with small user-specifiable overheads (e.g. 7%), and still provide hindsight log replay times orders of magnitude faster than restarting training from scratch.