Kalman Tiboldi - CTO & Founder at GemOne-TVH - Simplify or Die: Why Complexity Is the Real IT Challenge
Find out more:
Belgium 30-9-25 Squad Only Virtual english
Everyone has technical debt. The question is what to do with it. Refactor now? Wait for the next feature? Build on it and hope it holds? If you’re tired of debating whether to clean up or move on, this session helps you frame the debt conversation in a way that supports both speed and stability.
Read MorePoland 30-9-25 Physical
Zespoły i ludzie w dobie zmiany w sposobie dostarczania usług. Jak przechodzimy przez złożone zmiany technologiczne i organizacyjne? Czy transformacje w obszarach infrastruktury zmieniają wymagania na przywództwo i kompetencje zespołów.
Read MoreUK 30-9-25 Invitation Only Physical english
This exclusive Jefferson dinner, hosted by Mendix and CIONET, brings together a hand-picked group of senior executives and digital leaders from across the UK financial sector.
Read MoreBelgium 1-10-25 Invitation Only Physical english
In today’s rapidly evolving technological landscape, organisations face a critical challenge: how to modernise their legacy systems while managing costs, maintaining security, and ensuring operational continuity. This exclusive round table discussion brings together CIOs, CTOs, and Digital Leaders to explore best practices for legacy and lifecycle management. Key topics include extending the lifespan of existing systems, mitigating cyber security risks, optimising costs, and integrating legacy platforms with modern technologies. We’ll also delve into sustainability, workforce challenges, and future-proofing IT infrastructures to navigate technological obsolescence.
Read MoreBelgium 2-10-25 Squad Only Virtual english
Architecture doesn’t happen in diagrams. It happens in minds, seasoned, curious, sharp minds that understand how the pieces fit. But who’s shaping these minds? Who’s growing the next generation of digital architects? If you’re responsible for complex systems, and for the people designing them, this session is for you.
Read MoreGermany Sep 25, 2025 Country Members Physical english
Join us for an evening of thought-provoking discussions, real-world use cases, and insights from digital leaders, tech-pioneers and academics as we explore the dynamics of the human-machine relationship.
Read MoreNetherlands Sep 25, 2025 Invitation Only Physical dutch
Alleen op uitnodiging. How can organizations proactively ensure business continuity and IT resilience in 2025 amidst increasing disruptions?
Read MoreGermany Sep 29, 2025 Country Members Virtual german
Unsere Wissensaustausch-Briefings "AI & Data Knowledge Transfer" zielen darauf ab, CIDOs und Senior Digital Leaders mit praktischen Anwendungen und umsetzbaren Erkenntnissen von Fachexperten zu vermitteln, die den Weg für transformative digitale Reisen ebnen.
Read MoreGermany Sep 30, 2025 Country Members Physical german
Von der Theorie in die Praxis: Erst Business, dann Bier.Unstrukturierte Daten sind das verborgene Kapital vieler Unternehmen – und zugleich eine der größten Herausforderungen moderner IT. Gemeinsam mit CIOs, CDOs und CTOs der CIONET-Community sprechen wir zuerst bei WeWork München über Strategien, Technologien und echte Use Cases, um dieses Potenzial zu heben. Danach geht's weiter auf die Wiesn – für den offenen Austausch in bayerischer Atmosphäre.
Read MoreUK Sep 30, 2025 Invitation Only Physical english
This exclusive Jefferson dinner, hosted by Mendix and CIONET, brings together a hand-picked group of senior executives and digital leaders from across the UK financial sector.
Read MoreNetherlands Sep 30, 2025 Invitation Only Physical dutch
Alleen op uitnodiging. Mastering Digital Sovereignty Control, trust and resilience in an era of geopolitical and regulatory shifts. From the Network to the Data.
Read MoreGermany Sep 25, 2025 Country Members Physical english
Join us for an evening of thought-provoking discussions, real-world use cases, and insights from digital leaders, tech-pioneers and academics as we explore the dynamics of the human-machine relationship.
Read MoreNetherlands Sep 25, 2025 Invitation Only Physical dutch
Alleen op uitnodiging. How can organizations proactively ensure business continuity and IT resilience in 2025 amidst increasing disruptions?
Read MoreGermany Sep 29, 2025 Country Members Virtual german
Unsere Wissensaustausch-Briefings "AI & Data Knowledge Transfer" zielen darauf ab, CIDOs und Senior Digital Leaders mit praktischen Anwendungen und umsetzbaren Erkenntnissen von Fachexperten zu vermitteln, die den Weg für transformative digitale Reisen ebnen.
Read MoreGermany Sep 30, 2025 Country Members Physical german
Von der Theorie in die Praxis: Erst Business, dann Bier.Unstrukturierte Daten sind das verborgene Kapital vieler Unternehmen – und zugleich eine der größten Herausforderungen moderner IT. Gemeinsam mit CIOs, CDOs und CTOs der CIONET-Community sprechen wir zuerst bei WeWork München über Strategien, Technologien und echte Use Cases, um dieses Potenzial zu heben. Danach geht's weiter auf die Wiesn – für den offenen Austausch in bayerischer Atmosphäre.
Read MoreUK Sep 30, 2025 Invitation Only Physical english
This exclusive Jefferson dinner, hosted by Mendix and CIONET, brings together a hand-picked group of senior executives and digital leaders from across the UK financial sector.
Read MoreNetherlands Sep 30, 2025 Invitation Only Physical dutch
Alleen op uitnodiging. Mastering Digital Sovereignty Control, trust and resilience in an era of geopolitical and regulatory shifts. From the Network to the Data.
Read MoreHow Cohere is accelerating language model training with Google Cloud TPUs
Cohere is accelerating LLM training with Google Cloud TPUs to provide larger and more accurate LLMs to developers.
Machine Learning Engineer, Cohere
Sr. Product Manager
Over the past few years, advances in training large language models (LLMs) have moved natural language processing (NLP) from a bleeding-edge technology that few companies could access, to a powerful component of many common applications. From chatbots to content moderation to categorization, a general rule for NLP is that the larger the model, the greater the accuracy it’s able to achieve in understanding and generating language.
But in the quest to create larger and more powerful language models, scale has become a major challenge. Once a model becomes too large to fit on a single device, it requires distributed training strategies, which in turn require extensive compute resources with vast memory capacity and fast interconnects. You also need specialized algorithms to optimize the hardware and time resources.
Cohere engineers are working on solutions to this scaling challenge that have already yielded results. Cohere provides developers a platform for working with powerful LLMs without the infrastructure or deep ML expertise that such projects typically require. In a new technical paper, Scalable Training of Language Models using JAX pjit and TPUv4, engineers at Cohere demonstrate how their new FAX framework deployed on Google Cloud’s recently announced Cloud TPU v4 Pods addresses the challenges of scaling LLMs to hundreds of billions of parameters. Specifically, the report reveals breakthroughs in training efficiency that Cohere was able to achieve through tensor and data parallelism.
This framework aims to accelerate the research, development, and production of large language models with two significant improvements: scalability and rapid prototyping. Cohere will be able to improve its models by training larger ones more quickly, delivering better models to its customers faster. The framework also supports rapid prototyping of models that address specific objectives — for example, creating a generative model that powers customer-service chatbot — by experimenting and testing new ideas. The ability to switch back and forth among model types and optimize for different objectives will ultimately allow Cohere to offer models optimized for particular use cases.
The FAX framework relies heavily on the partitioned just-in-time compilation (pjit) feature of JAX, which abstracts the relationship between device and workload. This allows Cohere engineers to optimize efficiency, and performance by aligning devices and processes in the ideal configuration for the task at hand. Pjit works by compiling an arbitrary function into a single program (an XLA computation), that runs on multiple devices — even those residing on different hosts.
Cohere’s new solution also takes advantage of Google Cloud’s new TPU v4 Pods to perform tensor parallelism. which is more efficient than the earlier pipeline parallelism implementation. As the name suggests, the pipeline parallel approach uses accelerators in a linear fashion to scale a workload, like a single long assembly line. Accelerators must process each micro-batch of data before passing it along to the next one, and then run the backward pass in reverse order.
Tensor parallelism eliminates the accelerator idle time of pipeline parallelism, also known as the pipeline bubble. Tensor parallelism involves partitioning large tensors (mathematical arrays that define the relationship among multiple objects such as the words in a paragraph) across accelerators to perform computations at the same time on multiple devices. If pipeline parallelism is an ever-lengthening assembly line, tensor parallelism is a series of parallel assembly lines — one making the engine, the other the body, etc. — that simultaneously come together to form a complete car in a fraction of the time.
These computations are then collated, a process made practical thanks to Google Cloud TPU v4 VMs, which more than double the computational power of their v3 predecessors. The superior performance of v4 chips has enabled Cohere to iterate on ideas and validate them 1.7X faster in computation than before.
Aidan Gomez, CEO and co-founder, Cohere
As part of a multiyear technology partnership, Cohere leverages Google Cloud’s advanced AI and ML infrastructure to power its platform. Cohere develops and deploys its products on Cloud TPUs, Google Cloud’s custom-designed machine learning chips that are optimized for large-scale ML. Cohere’s recently announced their new model improvements and scalability by training an LLM using FAX on Google Cloud TPUs, and this model has demonstrated that transitioning from TPU v3 to TPU v4 has so far enabled them to achieve a total speedup of 1.7x . In addition to a significant performance boost, TPUs provide an excellent user experience with the new TPU VM architecture. Importantly, Google Cloud ensures that Cohere's state-of-the-art ML training is achieved with the highest standards of sustainability, powered by 90% carbon-free energy in the world's largest publicly available ML hub.
By adopting Cloud TPUs, Cohere is making LLM training faster, more economical, and more agile. This helps them provide larger and more accurate LLMs to developers, and put NLP technology in the hands of developers and businesses of all sizes.
To learn more about these LLM training advances, you can read the full paper, Scalable Training of Language Models using JAX pjit and TPUv4. To learn more about Cohere's best practices and AI principles, you can check this article co-authored with Open AI and AI 21 Labs.
470 Views 0 Likes Read More
Embark on a culinary journey through the ever-evolving world of digital leadership with our third edition of the CIONET Cookbook: Recipes for Digital Success. Unveiling the intricate trilemma faced by today’s Master Chefs, our trailblazing European CIOs address a challenge at the nexus of customer interests, digital transformation strategies, and IT modernisation. Their secret? Synchronising the gearing between customer, business, and technology to create a frictionless movement through the digital landscape.
The CIONET Cookbook uses the analogy of a five-star restaurant to explain the importance of optimally integrated technology, with the CIO as Master Chef. In order to provide the best service to its customers, a top restaurant must have the right atmosphere, an inviting menu, a well-equipped kitchen, talented and committed front-of-house and kitchen staff and smooth-running processes that ensure an enjoyable experience for diners.
As AI transitions from a specialised tool to an all-pervasive force, understanding its profound implications for our human lives is no longer merely academic but an urgent imperative for social well-being and strategic preparedness.
Geopolitical tensions have dominated the headlines for over two years now. In this context, the cyber threat landscape is also evolving rapidly. The protection and security of critical infrastructure – both physical and digital – is becoming increasingly important.
In our new app, our members connect with other digital leaders from around the world to find better solutions to their challenges.
Connect with digital leaders like you
Share ideas, best practices, and new resources
Experience inspiring and thought-provoking content and conversations you can’t find anywhere else
Make better, more well-informed decisions about the topics that are most important to you
Head of IT Germany and Regional Office Lead Munich, Microsoft
IT Director, IT Thinker
Vice president IT & Compliance, VisionaryRCM (A Carlyle Grp Company)
CIONET’s mission is to help IT executives become more at ease and above all more successful in their jobs. So they can do more than just keep up with change but ultimately define it. CIONET opens up a whole new universe of opportunities in IT management.
With the largest membership of corporate digital leaders across Europe, Latin America, US and Australia, CIONET has the expertise and pioneering vision to solve or address any IT management challenge.
From our local and global events, from our publications and research to our executive education programmes, everything we do is aimed at making sure digital leaders maximise their potential.
with digital leaders who share your interests, who face the same challenges, who care about the same topics.
stories, experiences, and ideas around our shared mission.
from our exclusive events, publications and research.
inspiration, thought-provoking conversations, expert perspectives and exclusive first-hand content each and every day
and make better, more well-informed decisions on how to lead your digital business.
your potential. Realise your ambitions.
You can either send us a registered handwritten letter explaining why you'd like to become a member or you can simply talk to us right here!
Would you like to know more about CIONET, membership or partnership opportunities? Do you have feedback or any other question? Send us a message!