Subscribe to the Machine Learning Engineer Newsletter

Receive curated articles, tutorials and blog posts from experienced Machine Learning professionals.

Issue #248
This 248 edition of the ML Engineer newsletter contains curated ML tutorials, OSS tools and AI events for our 45,000+  subscribers. You can access the Web Newsletter Homepage as well as the Linkedin Newsletter Homepage where you can find all previous editions πŸš€
If you like the content please support the newsletter by sharing with your friends via 🐦 Twitter,  πŸ’Ό Linkedin and  πŸ“• Facebook!
 This week in the ML Engineer:
Thank you for being part of over 45,000+ ML professionals and enthusiasts who receive weekly articles & tutorials on production ML & MLOps πŸ€– If you havent, you can join for free at ⭐
If you would like to suggest articles, ideas, papers, libraries, jobs, events or provide feedback just send us an email to! We have received a lot of great suggestions in the past, thank you very much for everyone's support!
The new Text-to-Song model from StabilityAI is a reminder for the exciting journey we are in the AI ecosystem πŸ€” Stability AI introduces a  latent diffusion model for audio generation, "Stable Audio". A model that accepts inputs as text description, audio duration, and start time, it addresses the challenge of producing varying audio lengths, such as full songs. The model leverages a variational autoencoder, a text encoder, and a U-Net-based conditioned diffusion model for efficient generation. Trained on over 800,000 audio files totaling 19,500 hours, Stable Audio promises faster inference times and enhanced controllability, with future releases set to include open-source models and training code.
Sebastian Raschka reminds us that quality over quantity and less is more is key when it comes to Large Language Model optimizations  πŸ’‘ In his most recent article, "Optimizing LLMs From a Dataset Perspective", he delves into optimizing Large Language Models by finetuning them with curated datasets. This basically emphasises the importance of instruction-based finetuning, contrasting human-created and LLM-generated datasets. It delves into the LIMA dataset's effectiveness, emphasizing quality over quantity, and offers a walkthrough for finetuning LLMs using the Lit-GPT repository.
Google contributes to the interesting discourse of transformer-based forecasting with the release of "TSMixer," an MLP architecture for time series forecasting that merges the strengths of univariate linear models with multivariate models. Unlike traditional Transformer architectures, TSMixer replaces attention mechanisms with linear layers, inspired by the computer vision MLP-Mixer method. In evaluations, TSMixer matched top univariate models on long-term forecasting benchmarks and showcased superior performance on the M5 retail dataset, emphasizing its potential in real-world applications and suggesting a reevaluation of the role of cross-variate information in forecasting.
Hallucination Detection for LLM-Based Abstractive Summaries. Eugene Yan shares another insightful article on LLMs that analyses the topic of summarization, and delves into evaluation challenges, discussing metrics like ROUGE, METEOR, BERTScore, MoverScore, ROUGE-C, and G-Eval. It also makes suggestions on key areas, such as being careful from being over-reliant on reference summaries, as well as to keep an eye on such as consistency in summaries.
Death by a thousand microservices πŸ₯· A great critique on the tech industry's over-reliance on microservices, highlighting inherent complexities and pitfalls. This article emphasises the importance of the ever long advice in the software engineering: "Don’t solve problems you don’t have". Although microservice architectures can be revolutionary, there is a time and a place for every design decisions, and many successful companies began with simpler monolithic architectures, suggesting that the blind adoption of microservices often leads to redundancy, decreased developer efficiency, and testing challenges. The article also outlines a trend towards returning to simpler architectures such as monolithic architectures, which urges companies to assess the genuine necessity of microservices based on their specific needs and scale.
Upcoming MLOps Events
The MLOps ecosystem continues to grow at break-neck speeds, making it ever harder for us as practitioners to stay up to date with relevant developments. A fantsatic way to keep on-top of relevant resources is through the great community and events that the MLOps and Production ML ecosystem offers. This is the reason why we have started curating a list of upcoming events in the space, which are outlined below.
Relevant upcoming MLOps conferences:
Open Source MLOps Tools
Check out the fast-growing ecosystem of production ML tools & frameworks at the github repository which has reached over 10,000 ⭐ github stars. We are currently looking for more libraries to add - if you know of any that are not listed, please let us know or feel free to add a PR. Four featured libraries in the GPU acceleration space are outlined below.
  • Kompute - Blazing fast, lightweight and mobile phone-enabled GPU compute framework optimized for advanced  data processing usecases.
  • CuPy - An implementation of NumPy-compatible multi-dimensional array on CUDA. CuPy consists of the core multi-dimensional array class, cupy.ndarray, and many functions on it.
  • Jax - Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
  • CuDF - Built based on the Apache Arrow columnar memory format, cuDF is a GPU DataFrame library for loading, joining, aggregating, filtering, and otherwise manipulating data.
If you know of any open source and open community events that are not listed do give us a heads up so we can add them!
As AI systems become more prevalent in society, we face bigger and tougher societal challenges. We have seen a large number of resources that aim to takle these challenges in the form of AI Guidelines, Principles, Ethics Frameworks, etc, however there are so many resources it is hard to navigate. Because of this we started an Open Source initiative that aims to map the ecosystem to make it simpler to navigate. You can find multiple principles in the repo - some examples include the following:
  • MLSecOps Top 10 Vulnerabilities - This is an initiative that aims to further the field of machine learning security by identifying the top 10 most common vulnerabiliites in the machine learning lifecycle as well as best practices.
  • AI & Machine Learning 8 principles for Responsible ML - The Institute for Ethical AI & Machine Learning has put together 8 principles for responsible machine learning that are to be adopted by individuals and delivery teams designing, building and operating machine learning systems.
  • An Evaluation of Guidelines - The Ethics of Ethics; A research paper that analyses multiple Ethics principles.
  • ACM's Code of Ethics and Professional Conduct - This is the code of ethics that has been put together in 1992 by the Association for Computer Machinery and updated in 2018.
If you know of any guidelines that are not in the "Awesome AI Guidelines" list, please do give us a heads up or feel free to add a pull request!
Β© 2018 The Institute for Ethical AI & Machine Learning