Simon Holm
@simonholm.bsky.social
I'm tinkering with code and circuits. The woodland is my go-to retreat
Reposted by Simon Holm
November 17, 2025 at 4:20 PM
Reposted by Simon Holm
If you some excellent wildlife photography on BlueSky, I recommend adding the feed for #BirdOfTheDay.
November 17, 2025 at 4:12 PM
If you some excellent wildlife photography on BlueSky, I recommend adding the feed for #BirdOfTheDay.
Reposted by Simon Holm
The little terminal that could 💖 so proud to be a part of this amazing community awarding Windows Terminal with 100K GitHub stars 🌟
September 9, 2025 at 6:34 PM
The little terminal that could 💖 so proud to be a part of this amazing community awarding Windows Terminal with 100K GitHub stars 🌟
Reposted by Simon Holm
v1.108 is rolling out today 🚚
Now live, at long last: Bookmarks, aka Saved Posts. For all those posts you'll definitely plan to come back to!
Update the app and give it a try. The button is right down there 👇
Now live, at long last: Bookmarks, aka Saved Posts. For all those posts you'll definitely plan to come back to!
Update the app and give it a try. The button is right down there 👇
September 8, 2025 at 6:24 PM
v1.108 is rolling out today 🚚
Now live, at long last: Bookmarks, aka Saved Posts. For all those posts you'll definitely plan to come back to!
Update the app and give it a try. The button is right down there 👇
Now live, at long last: Bookmarks, aka Saved Posts. For all those posts you'll definitely plan to come back to!
Update the app and give it a try. The button is right down there 👇
Reposted by Simon Holm
In the coming weeks, we're asking for feedback on our Community Guidelines. Your input will help ensure our Guidelines serve our users and are ready to support Bluesky's future.
To learn how to contribute, check out our blog post: bsky.social/about/blog/0...
To learn how to contribute, check out our blog post: bsky.social/about/blog/0...
Updated Terms and Policies - Bluesky
We’re updating the language in our terms and policies to better explain our approach and provide more detail.
bsky.social
August 14, 2025 at 4:32 PM
In the coming weeks, we're asking for feedback on our Community Guidelines. Your input will help ensure our Guidelines serve our users and are ready to support Bluesky's future.
To learn how to contribute, check out our blog post: bsky.social/about/blog/0...
To learn how to contribute, check out our blog post: bsky.social/about/blog/0...
Reposted by Simon Holm
I'm very pleased to share our latest study:
‘Emergence of Language in the Developing Brain’,
by L Evanson, P Bourdillon et al:
- Paper: ai.meta.com/research/pub...
- Blog: ai.meta.com/blog/meta-fa...
- Thread below 👇
‘Emergence of Language in the Developing Brain’,
by L Evanson, P Bourdillon et al:
- Paper: ai.meta.com/research/pub...
- Blog: ai.meta.com/blog/meta-fa...
- Thread below 👇
May 15, 2025 at 4:00 PM
I'm very pleased to share our latest study:
‘Emergence of Language in the Developing Brain’,
by L Evanson, P Bourdillon et al:
- Paper: ai.meta.com/research/pub...
- Blog: ai.meta.com/blog/meta-fa...
- Thread below 👇
‘Emergence of Language in the Developing Brain’,
by L Evanson, P Bourdillon et al:
- Paper: ai.meta.com/research/pub...
- Blog: ai.meta.com/blog/meta-fa...
- Thread below 👇
Reposted by Simon Holm
ChatGPT Switches Default Model to GPT-4.1 Mini for All Users, Subscribers Get GPT-4.1
OpenAI has integrated its GPT-4.1 model into ChatGPT, making it available to Plus, Pro, and Team subscribers. The company announced the rollout on social media, highlighting the model's str…
#chatgpt #gpt #openai
OpenAI has integrated its GPT-4.1 model into ChatGPT, making it available to Plus, Pro, and Team subscribers. The company announced the rollout on social media, highlighting the model's str…
#chatgpt #gpt #openai
ChatGPT Switches Default Model to GPT-4.1 Mini for All Users, Subscribers Get GPT-4.1
OpenAI has integrated its GPT-4.1 model into ChatGPT, making it available to Plus, Pro, and Team subscribers. The company announced the rollout on social media, highlighting the model's strengths in coding tasks and instruction following. The GPT-4.1 model will also be available to Enterprise and Education users in the coming weeks. Alongside the premium upgrades, OpenAI is replacing GPT-4o mini with GPT-4.1 mini as the default model for all users. This change will affect all users, including those with free accounts, and will also impact the ChatGPT extension for Apple Intelligence. The new models support a one million context token window, which is a significant increase from GPT-4o's 128,000-token limit. OpenAI claims that the speed improvements in GPT-4.1 make it more suitable for everyday coding tasks. The deployment of the new models comes a few weeks after they were released to developers, and they have undergone standard safety evaluations. The integration of GPT-4.1 into ChatGPT is expected to enhance the overall user experience, especially for those who use the platform for coding and instruction-based tasks. Overall, the update is a significant improvement to the ChatGPT platform, and users can expect to see the benefits of the new models in the coming weeks.
www.macrumors.com
May 16, 2025 at 3:18 AM
Reposted by Simon Holm
Harvard, the world’s richest university, may have enough money to survive a battle with President Trump.
But if Trump wins, Harvard won’t be the same.
But if Trump wins, Harvard won’t be the same.
Can Harvard Withstand Trump’s Financial Attack?
The world’s richest university may have enough money to survive a battle with the most powerful man in the world. But if Trump wins, Harvard won’t be the same.
www.nytimes.com
April 22, 2025 at 12:47 PM
Harvard, the world’s richest university, may have enough money to survive a battle with President Trump.
But if Trump wins, Harvard won’t be the same.
But if Trump wins, Harvard won’t be the same.
Reposted by Simon Holm
Help shape the EU guidelines on general-purpose AI (GPAI) models under the AI Act.
We invite all stakeholders and citizens to share feedback by 22 May in a targeted consultation.
More info: europa.eu/!qt4Ypy
#EUHaveYourSay
We invite all stakeholders and citizens to share feedback by 22 May in a targeted consultation.
More info: europa.eu/!qt4Ypy
#EUHaveYourSay
April 22, 2025 at 11:04 AM
Help shape the EU guidelines on general-purpose AI (GPAI) models under the AI Act.
We invite all stakeholders and citizens to share feedback by 22 May in a targeted consultation.
More info: europa.eu/!qt4Ypy
#EUHaveYourSay
We invite all stakeholders and citizens to share feedback by 22 May in a targeted consultation.
More info: europa.eu/!qt4Ypy
#EUHaveYourSay
Reposted by Simon Holm
scoop: OpenAI gets ready to launch GPT-4.1. OpenAI is getting ready to launch a collection of new models and features, including o3 and o4 mini www.theverge.com/news/646458/...
OpenAI gets ready to launch GPT-4.1
o4 mini and o3 might also debut next week
www.theverge.com
April 10, 2025 at 3:47 PM
scoop: OpenAI gets ready to launch GPT-4.1. OpenAI is getting ready to launch a collection of new models and features, including o3 and o4 mini www.theverge.com/news/646458/...
Reposted by Simon Holm
👿 MCP is all fun, until you add this one malicious MCP server and forget about it.
We have discovered a critical flaw in the widely-used Model Context Protocol (MCP) that enables a new form of LLM attack we term 'Tool Poisoning'.
Leaks SSH key, API keys, etc.
Details below 👇
We have discovered a critical flaw in the widely-used Model Context Protocol (MCP) that enables a new form of LLM attack we term 'Tool Poisoning'.
Leaks SSH key, API keys, etc.
Details below 👇
April 3, 2025 at 7:47 AM
👿 MCP is all fun, until you add this one malicious MCP server and forget about it.
We have discovered a critical flaw in the widely-used Model Context Protocol (MCP) that enables a new form of LLM attack we term 'Tool Poisoning'.
Leaks SSH key, API keys, etc.
Details below 👇
We have discovered a critical flaw in the widely-used Model Context Protocol (MCP) that enables a new form of LLM attack we term 'Tool Poisoning'.
Leaks SSH key, API keys, etc.
Details below 👇
Reposted by Simon Holm
> Almost half of C-suite executives said in a recent survey that AI adoption is “tearing their company apart” as a rift emerges between leadership and the employees adopting such tools.
leaddev.com/culture/ai-c...
leaddev.com/culture/ai-c...
AI coding mandates are driving developers to the brink
Under pressure to embrace AI, developers are growing frustrated by misguided mandates and are left to clean up any collateral damage inflicted on their codebase.
leaddev.com
April 8, 2025 at 6:36 PM
> Almost half of C-suite executives said in a recent survey that AI adoption is “tearing their company apart” as a rift emerges between leadership and the employees adopting such tools.
leaddev.com/culture/ai-c...
leaddev.com/culture/ai-c...
Reposted by Simon Holm
I received a heartbreaking email today: A paper in a special issue I’m editing is being retracted because one of its authors is afraid of losing their job and their legal status in the U.S. if they publish a scientific study on evolution. Yes, on evolution, nature's engine of diversity.
March 29, 2025 at 11:52 PM
I received a heartbreaking email today: A paper in a special issue I’m editing is being retracted because one of its authors is afraid of losing their job and their legal status in the U.S. if they publish a scientific study on evolution. Yes, on evolution, nature's engine of diversity.
Reposted by Simon Holm
Microsoft's Playwright, that enables reliable end-to-end testing for modern web apps ( playwright.dev ) built a Playwright MCP ( github.com/microsoft/pl... ) that provides browser automation using Playwright. This enables LLMs to interact with web pages through structured accessibility snapshots.
Fast and reliable end-to-end testing for modern web apps | Playwright
Cross-browser end-to-end testing for modern web apps
playwright.dev
March 25, 2025 at 3:46 AM
Microsoft's Playwright, that enables reliable end-to-end testing for modern web apps ( playwright.dev ) built a Playwright MCP ( github.com/microsoft/pl... ) that provides browser automation using Playwright. This enables LLMs to interact with web pages through structured accessibility snapshots.
Reposted by Simon Holm
🚀 We just dropped SmolDocling: a 256M open-source vision LM for complete document OCR! 📄✨
Lightning fast, process a page in 0.35 sec on consumer GPU using < 500MB VRAM ⚡
SOTA in document conversion, beating every competing model we tested (including models 27x more params) 🤯
But how? 🧶⬇️
Lightning fast, process a page in 0.35 sec on consumer GPU using < 500MB VRAM ⚡
SOTA in document conversion, beating every competing model we tested (including models 27x more params) 🤯
But how? 🧶⬇️
March 17, 2025 at 3:53 PM
🚀 We just dropped SmolDocling: a 256M open-source vision LM for complete document OCR! 📄✨
Lightning fast, process a page in 0.35 sec on consumer GPU using < 500MB VRAM ⚡
SOTA in document conversion, beating every competing model we tested (including models 27x more params) 🤯
But how? 🧶⬇️
Lightning fast, process a page in 0.35 sec on consumer GPU using < 500MB VRAM ⚡
SOTA in document conversion, beating every competing model we tested (including models 27x more params) 🤯
But how? 🧶⬇️
Reposted by Simon Holm
ChatGPT’s latest update closes the gap against Gemini on Android
You can now set ChatGPT as the default assistant on Android, letting you invoke it through buttons and gestures.
#chatgpt #geminiai #gpt
You can now set ChatGPT as the default assistant on Android, letting you invoke it through buttons and gestures.
#chatgpt #geminiai #gpt
ChatGPT’s latest update closes the gap against Gemini on Android
You can now set ChatGPT as the default assistant on Android, letting you invoke it through buttons and gestures.
www.androidauthority.com
March 15, 2025 at 5:46 AM
Reposted by Simon Holm
Gemini will now also remember free users’ preferences
Google expands Gemini Advanced's Saved Info feature to free users.
#ai #gemini #geminiai
Google expands Gemini Advanced's Saved Info feature to free users.
#ai #gemini #geminiai
Gemini will now also remember free users’ preferences
Google expands Gemini Advanced's Saved Info feature to free users.
www.androidauthority.com
March 4, 2025 at 6:06 AM
Reposted by Simon Holm
Actual leadership:
UK PM Starmer to Zelensky:
"You are very, very welcome here in Downing Street. As you heard from the cheers on the street outside, you have full backing across the United Kingdom. We will stand with you and Ukraine for as long as it may take."
"You are very, very welcome here in Downing Street. As you heard from the cheers on the street outside, you have full backing across the United Kingdom. We will stand with you and Ukraine for as long as it may take."
March 1, 2025 at 5:46 PM
Actual leadership:
Reposted by Simon Holm
Today, we’re launching new algorithms that enable 10x longer context lengths & 90% less VRAM for training Reasoning Models (GRPO).
Using Unsloth, you can now train your own reasoning model with just 5GB VRAM for Qwen2.5-1.5B with no accuracy loss.
Blog: unsloth.ai/blog/grpo
Using Unsloth, you can now train your own reasoning model with just 5GB VRAM for Qwen2.5-1.5B with no accuracy loss.
Blog: unsloth.ai/blog/grpo
February 20, 2025 at 6:41 PM
Today, we’re launching new algorithms that enable 10x longer context lengths & 90% less VRAM for training Reasoning Models (GRPO).
Using Unsloth, you can now train your own reasoning model with just 5GB VRAM for Qwen2.5-1.5B with no accuracy loss.
Blog: unsloth.ai/blog/grpo
Using Unsloth, you can now train your own reasoning model with just 5GB VRAM for Qwen2.5-1.5B with no accuracy loss.
Blog: unsloth.ai/blog/grpo
Reposted by Simon Holm
The Agent Leaderboard on Huggingface🔥
Evaluate LLM's ability to utilize tools in complex scenarios. Understand how AI agents perform in real-world business scenarios. Stunning leaderboard created with Gradio 5 😍
Evaluate LLM's ability to utilize tools in complex scenarios. Understand how AI agents perform in real-world business scenarios. Stunning leaderboard created with Gradio 5 😍
February 13, 2025 at 6:11 AM
The Agent Leaderboard on Huggingface🔥
Evaluate LLM's ability to utilize tools in complex scenarios. Understand how AI agents perform in real-world business scenarios. Stunning leaderboard created with Gradio 5 😍
Evaluate LLM's ability to utilize tools in complex scenarios. Understand how AI agents perform in real-world business scenarios. Stunning leaderboard created with Gradio 5 😍
Reposted by Simon Holm
dani-garcia / vaultwarden: Unofficial Bitwarden compatible server written in Rust, formerly known as bitwarden_rs ★41636 https://github.com/dani-garcia/vaultwarden
dani-garcia / vaultwarden
Unofficial Bitwarden compatible server written in Rust, formerly known as bitwarden_rs
github.com
February 11, 2025 at 8:21 AM
dani-garcia / vaultwarden: Unofficial Bitwarden compatible server written in Rust, formerly known as bitwarden_rs ★41636 https://github.com/dani-garcia/vaultwarden
Reposted by Simon Holm
GitHub repo: github.com/fdarkaou/op...
GitHub - fdarkaou/open-deep-research: An open-source alternative to OpenAI and Gemini's deep research.
An open-source alternative to OpenAI and Gemini's deep research. - fdarkaou/open-deep-research
github.com
February 10, 2025 at 8:01 AM
GitHub repo: github.com/fdarkaou/op...
Reposted by Simon Holm
OpenAI's Deep Research costs $200/month.
I made a beautiful, free, and open-source version that's just as powerful.
Knowledge shouldn't be behind a paywall 👇
I made a beautiful, free, and open-source version that's just as powerful.
Knowledge shouldn't be behind a paywall 👇
February 10, 2025 at 8:01 AM
OpenAI's Deep Research costs $200/month.
I made a beautiful, free, and open-source version that's just as powerful.
Knowledge shouldn't be behind a paywall 👇
I made a beautiful, free, and open-source version that's just as powerful.
Knowledge shouldn't be behind a paywall 👇