2.5 Admins 309: Orbital Sauron

2.5 Admins 309: Orbital Sauron

Video by The Late Night Linux Family via YouTube
2.5 Admins 309: Orbital Sauron

Support us on Patreon and get an ad-free RSS feed with some early episodes. https://www.patreon.com/LateNightLinux

A new HTTP caching standard looks promising but might struggle to get adopted, a terrible space mirror idea is actually happening, cars will spy on us all, and picking a VPN for use with an IP that doesn’t change very often.

https://2.5admins.com/2-5-admins-309/

Source

Trino contributor call 2026-07-22

Trino contributor call 2026-07-22

Video by Trino via YouTube
Trino contributor call 2026-07-22

– 00:00 Trino 483 release and SQL improvements
– 04:06 New Trino web UI becomes the default
– 05:45 Production memory fixes
– 09:02 Anonymous S3 access
– 11:20 CLI themes and Trino packages
– 12:41 Trino Gateway 21 and the spooled client protocol
– 13:54 Client releases and build improvements
– 17:20 Community call to action

More details available in the meeting minutes at
https://github.com/trinodb/trino/wiki/Contributor-meetings

Call organized and hosted by Manfred Moser. Sponsor him at
https://github.com/sponsors/mosabua to support this and other open source initiatives.

Source

Easy Unit Conversions on Your iPhone! #shorts

Easy Unit Conversions on Your iPhone! #shorts

Video by TWiT Tech Podcast Network via YouTube
Easy Unit Conversions on Your iPhone! #shorts

Go beyond basic math! Discover your device’s built-in calculator, featuring over 110 unit conversions (temp, volume, energy, and more!) and 105 currency options. Effortless conversions, no manual input needed. #CalculatorTips #TechHacks #Productivity #LifeHacks

Source

The Next AI Breakthrough Isn’t Smarter Models

The Next AI Breakthrough Isn't Smarter Models

Video by Rasa via YouTube
The Next AI Breakthrough Isn't Smarter Models

🤖 AI doesn’t need to be *smarter.*

It needs to know **you** better.

Brian Alvey, CTO of WordPress VIP, believes the next breakthrough in AI won’t be answering more questions—it’s understanding your context.

Instead of simply telling you the weather, imagine an AI that knows where you’re headed, predicts what you’ll need, and helps you make better decisions before you even ask.

🌤️ "It’s 80° here…"
❄️ "…but you’re driving into San Francisco, where it’s 25 degrees colder. Grab a jacket."

That’s the kind of AI people will actually pay for.

The future of enterprise AI isn’t just better models—it’s more personalized, context-aware experiences.

🎧 Watch the full conversation with Lauren Goerz and Brian Alvey on *The Dialogue Architects*.

#ArtificialIntelligence #EnterpriseAI #ConversationalAI #WordPressVIP #AIProducts

Source

Open Source Digest: AI, Community, and Tools

Open Source Digest: AI, Community, and Tools

Community & Collaboration Social Coworking & Office Hours – SORTEE introduces social coworking with office hours. (Open Source Events) Vale and Text Linting – Collaborative office hours focused on Vale prose linting. (Open Source Quality) AI is Community Driven – The future of AI hinges on open, community‑driven development. (AI & Open Source) Wikimedia Communities … Read more

Open Source AI: Global Clash and Opportunity

Open Source AI: Global Clash and Opportunity

The Open Source AI Battle: What’s at Stake This week, the open-source AI world is buzzing with a clash of titans. China has released massive open models, Nvidia’s CEO is urging the US to embrace them, and a rogue hacking agent from OpenAI has raised eyebrows. For open-source enthusiasts and developers, these stories highlight a … Read more

EU Digital Sovereignty, RL Agents, and Open Source Career Paths

EU Digital Sovereignty, RL Agents, and Open Source Career Paths

EuroCommons TechSprint: Building Europe’s Digital Sovereignty The EuroCommons TechSprint in Paris brought together organizations like Caisse des DĂ©pĂ´ts, TOSIT, OW2, and XWiki to collaborate on reducing dependency on proprietary software and building trusted open source alternatives. Key figures emphasized that EuroCommons is about collaboration across organizations to create sustainable digital infrastructure that gives users control … Read more

A day at EuroCommons TechSprint

A day at EuroCommons TechSprint

Video by OpenProject | Open Source Project Management via YouTube
A day at EuroCommons TechSprint

The EuroCommons TechSprint in Paris brought together organizations working toward a stronger, more sovereign European digital ecosystem.

It was a pleasure to connect with inspiring partners and communities, including Caisse des Dépôts, TOSIT, OW2, XWiki, and many others who share a joint vision:
reducing dependency on proprietary software and building trusted open source alternatives together.

In this video, hear from Hugues de Maussion, Aymeril Hoang, Gilles Viton, Simon Clavier, Florian Caringi, Pauline Bessoles, Rosanna Sibora and Wieland Lindenthal as they share why collaboration is essential to strengthening Europe’s digital sovereignty.

A special thank you to Dicle Naz Tohumcu for creating this video and to Christophe Bliard for helping make the EuroCommons TechSprint such a success.

EuroCommons is about more than technology. It’s about collaboration across organizations to create sustainable digital infrastructure that gives users control over their data, their tools, and their future. We’re excited to continue contributing to this growing ecosystem and helping build open, trusted digital infrastructure for Europe.

🎥 Watch Europe’s open source ecosystem come together.

#EuroCommons #OpenProject #XWiki #OpenSource #DigitalSovereignty #TOSIT #CaissedesDépôts

🎵
Code: R0F2DWYDCUS2QLFS

Source

Training Agents 3: Reinforcement Learning

Training Agents 3: Reinforcement Learning

Video by Hugging Face via YouTube
Training Agents 3: Reinforcement Learning

Training Agents, Session 3: from a fine-tuned baseline to reinforcement learning with GRPO.

In this live session, we’ll move past imitation and start training on outcomes. We’ll take the model from the earlier sessions and run Group Relative Policy Optimization: sample multiple completions per prompt, score them with a reward function, and use the group’s relative scores as the training signal — no value model, no reward model, no critic to train. Everything runs in TRL.

What we’ll cover:
– Why GRPO after SFT, and what RL adds that imitation can’t
– How GRPO works: group sampling, relative advantages, and why that replaces a critic
– Writing verifiable reward functions for agent tasks
– Running TRL GRPO on Hugging Face Jobs
– Reading the training curves: reward, KL, and completion length
– Common failure modes: reward hacking, collapse, and length gaming

Repo: https://github.com/burtenshaw/training-agents

This is part of the Training Agents series: using coding agents to design, run, monitor, and review post-training experiments, while training models to become better agents.

Source