Issue #271 - The ML Engineer ??

Thank you for being part of over 60,000+ ML professionals and enthusiasts who receive weekly articles & tutorials on Machine Learning & MLOps ?? You can join the newsletter for free at https://ethical.institute/mle.html ?

If you like the content please support the newsletter by sharing with your friends via ?? Email, ?? Twitter , ?? Linkedin and ?? Facebook ! If you've come across this you can join the newsletter?asdf ?for free at https://ethical.institute/mle.html ? ?

If you are a Machine Learning Practitioner looking for an interesting opportunity, I'm currently hiring for a few roles including Senior Principal / Head of Product , Data Scientist Marketing Intelligence , Data Engineer Machine Learning & Data Engineer Data Platform - do check it out and do feel free to share broadly!

This week in Machine Learning:

• Salesforce Forecasting Foundation Model
• Google DeepMind Gemma.cpp
• Stanford Privacy in the AI Era
• Graph Neural Nets at Linkedin
• GPT in 60 Lines of NumPy
• Upcoming MLOps Events
• Open Source ML Frameworks
• Awesome AI Guidelines to check out this week
• + more ??

Salesforce FC Foundation Model

Salesforce enters the competition with their latest foundation model for Forecasting, following Google's announcement of their foundation model TimesFM : Salesforce introduces MOIRAI, a groundbreaking Large Time Series Model for universal time series forecasting. This new foundation model has been trained on the extensive Large-scale Open Time Series Archive with over 27 billion observations across nine domains. MOIRAI addresses key challenges in time series forecasting, including cross-frequency learning, multivariate data handling, and flexible distributional adaptation, through innovative approaches like multiple input/output projection layers, Any-variate Attention, and a mixture of parametric distributions. MOIRAI sets a new benchmark in the field demonstrating superior zero-shot forecasting capabilities compared to traditional full-shot models, offering a scalable and adaptable solution that significantly advances the potential for universal forecasting models in practical applications. The commitment to open-source the model weights, code, and LOTSA dataset further is a great contribution to the community, highlighting its potential impact on the industry, enabling widespread adoption and innovation in time series forecasting.

Google DeepMind Gemma.cpp

As the world continues to make waves from Google DeepMind's open source release of Gemma, the open source community already released an optimised C++ engine through "gemma.cpp": Following the steps of LLaMa.cpp, we now see the release of Gemma.cpp, bridging the gap between deployment-oriented runtimes and Python ML frameworks. This project leverages SIMD optimizations for the Gemma 2B and 7B models, including its large vocabulary and unique features like RMSNorm normalization and GeGLU activations. Open source contributions are key to unlock opportunities such as the finetuning from Sebastian Raschka , as well as important scrutinisation especially following the controversy surrounding Google's Gemini AI when skewing racial representation in image generation.

Stanford Privacy in the AI Era

Stanford has releases a whitepaper proposing the ecosystem to rethink privacy in the AI era: This whitepaper addresses the critical intersection of privacy, data protection, and artificial intelligence, highlighting the challenges and risks posed by the current and future landscape of AI development. It underscores the insufficiency of existing privacy laws to manage the escalating demand for data by AI systems, which not only threatens individual privacy but also poses broader societal risks. The paper advocates for a paradigm shift towards more stringent data collection norms, enhanced transparency and accountability throughout the AI data supply chain, and the development of new governance mechanisms to empower individuals in managing their data. Finally, the urgency for policymakers and stakeholders to adapt and enforce regulations that safeguard privacy while fostering responsible AI innovation, arguing that the future of AI and privacy is not predetermined but can be shaped by deliberate and thoughtful action.

Graph Neural Nets at Linkedin

Linkedin Research showcases their organisational adoption and learnings using graph neural network across their social network ecosystem: "LiGNN - Graph Neural Networks at LinkedIn" presents a comprehensive framework for deploying large-scale Graph Neural Networks within LinkedIn's ecosystem, addressing challenges unique to GNN training at scale, managing diverse entities, handling cold starts, and adapting to dynamic systems. They leverage a suite of novel techniques - ie. temporal graph architectures, graph densification, and efficient multi-hop neighbor sampling - in order to achieve significant improvements in LinkedIn's key metrics across various domains such as job applications, ads click-through rates, and user engagement. The deployment of LiGNN leverages adaptive sampling, data batching optimizations, and specialized infrastructure to accelerate training by 7x, demonstrating the practical applicability and effectiveness of GNNs in enhancing recommendation systems and user interaction on LinkedIn's platform. This work not only showcases the tangible benefits of applying GNNs at scale but also provides valuable insights and methodologies for production machine learning practitioners looking to harness the power of graph neural networks in large-scale applications.

GPT in 60 Lines of NumPy

Let's write GPT in 60 Lines of NumPy: What better way to learn a concept than implementing it ourselves; this is a great simplified yet complete introduction to the GPT architecture. In this case this leverages the trained GPT-2 model weights released by OpenAI, the implementation is capable of generating text, demonstrating the core functionalities of GPT models. This implementation intentionally omits many advanced features to maintain simplicity and understandability. The resource serves as a practical guide for machine learning practitioners interested in the inner workings of GPT models, providing insights into the model's architecture, including token and positional embeddings, the decoder stack, and the projection to the vocabulary layer.

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.

Upcoming conferences where we're speaking:

• AI in Production 2024 @ 15th February @ Virtual
• WeAreDevelopers 2024 @ 17th July @ Germany
• Data Intelligence Day 2024 @ 9th April @ Germany

Other upcoming MLOps conferences in 2024: ?

• FOSDEM (AI & HPC) - 3rd February @ Belgium
• State of the Open (Data) - 6th February @ UK
• World AI Cannes - 8th February @ France
• NVIDIA GTC - 17th March @ USA
• KubeCon Europe (AI) - 18th March @ France
• MLConf - 28th March @ USA
• PyData & PyCon DE - 17th April @ Germany
• MLSys - 13th May @ USA
• ODSC West - 29th October @ USA
• Data & AI Summit - 10th June @ USA
• AI Summit London - 12th June @ UK
• World Summit AI - 9th October @ Neatherlands

In case you missed our talks:

• Responsible AI Workshop Keynote - NeurIPS 2021
• Practical Guide to ML Explainability - PyCon London
• ML Monitoring: Outliers, Drift, XAI - PyCon Keynote
• Metadata for E2E MLOps - Kubecon NA 2022
• ML Performance Evaluation at Scale - KubeCon Eur 2021
• Industry Strength LLMs - PyData Global 2022
• ML Security Workshop Keynote - NeurIPS 2022

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!

OSS: Policy & Guidelines

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 !

About us

The Institute for Ethical AI & Machine Learning is a European research centre that carries out world-class research into responsible machine learning.

Check out our website