Mark Pors 🦖

5.9K posts

Mark Pors 🦖 banner
Mark Pors 🦖

Mark Pors 🦖

@pors

Building https://t.co/pRJXNjcrwa. Co-founder and CTO of Watchmouse (acquired by CA Technologies (acquired by Broadcom)).

The Hague, Netherlands Se unió Ekim 2007
1.8K Siguiendo1.2K Seguidores
Tweet fijado
Mark Pors 🦖
Mark Pors 🦖@pors·
Most neural nets are still based on the model of a neuron as proposed in the 1950's: u = activation(w·x + b) In a new paper, researchers propose a more accurate model of a biological brain neuron and found that it has quite a few advantages, like needing less training data.
Mark Pors 🦖 tweet media
English
33
123
1.3K
91.1K
Mark Pors 🦖 retuiteado
clem 🤗
clem 🤗@ClementDelangue·
Let’s face it: after-the-fact API guardrails are not the right safety tool for frontier models. They don’t make dangerous capabilities disappear. They just hide them behind a brittle interface that can be easily jailbroken. A better safety agenda: - don’t train models for very high-risk capabilities without strong evals, justification, and containment - use staged release, as pioneered by @IreneSolaiman, from trusted testers to broader access, and open release for transparency and accountability - massively support open-source AI so the gap between players does not become so large that a few closed labs and governments end up with overwhelming capabilities and power over everyone else - enable independent evaluation instead of asking everyone to trust a black-box API - give law enforcement, courts, regulators, auditors, journalists, and civil society strong AI tools to detect, investigate, and hold accountable unlawful uses of AI Safety means transparency, staged deployment, distributed power, and making sure democratic institutions can actually enforce the law.
English
18
9
109
9.7K
Mark Pors 🦖
Mark Pors 🦖@pors·
Researchers ask LLMs probability questions and found that by making the prompts leading the performance fell by as much as 34%. Paperzilla found this paper for me in my AI4Math feed. Summary and links to source: paperzilla.ai/digest/b19f502…
English
0
0
1
20
Mark Pors 🦖
Mark Pors 🦖@pors·
We have all noticed that asking a leading question to an LLM influences the answer. The answer tends to go into the direction that has your preference even if it is further from the actual truth. Interesting paper about this 👇
Mark Pors 🦖 tweet media
English
1
1
1
40
Mark Pors 🦖
Mark Pors 🦖@pors·
@mitsuhiko These generalizing idiots really piss me off. They obviously do it to get any form of response, but still.
English
0
0
2
554
Armin Ronacher ⇌
Armin Ronacher ⇌@mitsuhiko·
Why carry a water bottle if there is free water everywhere? x.com/ClintFiore/sta…
Armin Ronacher ⇌ tweet media
Clint Fiore 🦬 DM for Biz Deals@ClintFiore

Is everyone in Europe dehydrated? Nobody carries water bottles. Water is expensive everywhere and not free at most restaurants. Bathrooms are scarce and often cost money to use. My theory is everyone is ok with this bleak situation because they’re all massively dehydrated, rarely pee, and don’t know the glory of what 1+ gallons of water consumed per day feels like to the human body.

English
17
4
234
28.7K
Mark Pors 🦖
Mark Pors 🦖@pors·
@ClintFiore "Europe". In Greece a 1/2L bottle mineral water costs 50 cents. All these idiots bashing on "Europe" when they clearly have no idea. (no airco's in Europe either right?)
English
1
0
1
196
Clint Fiore 🦬 DM for Biz Deals
Clint Fiore 🦬 DM for Biz Deals@ClintFiore·
Is everyone in Europe dehydrated? Nobody carries water bottles. Water is expensive everywhere and not free at most restaurants. Bathrooms are scarce and often cost money to use. My theory is everyone is ok with this bleak situation because they’re all massively dehydrated, rarely pee, and don’t know the glory of what 1+ gallons of water consumed per day feels like to the human body.
English
2.3K
60
1.9K
1.9M
Mark Pors 🦖
Mark Pors 🦖@pors·
@scaling01 Unless you are actually coding daily. You would see that Openai >> Anthropic
English
1
0
9
345
Lisan al Gaib
Lisan al Gaib@scaling01·
believe it or not, but Anthropic, OpenAI and Google are all in different categories now Anthropic > OpenAI >> Google, SpaceX AI, DeepSeek, Moonshot, ZAI, Meta
Lisan al Gaib@scaling01

Noam Shazeer - co-author of the Transformer, T5, and Switch Transformer papers, and one of the pioneers of sparse MoE models - is leaving his VP Engineering / Gemini co-lead role at Google DeepMind to join OpenAI This is likely the most significant AI talent move of the year. It makes you wonder what’s going on at Google

English
64
26
845
125.9K
CJ Zafir
CJ Zafir@cjzafir·
Codex 5.5 at 4am is much smarter than Codex at 1pm EST. Just try it.
English
59
11
590
44K
clem 🤗
clem 🤗@ClementDelangue·
Another competitor bites the dust (ok it wasn’t really a direct competitor yet but could have become one overtime!)
clem 🤗 tweet media
English
28
19
353
43.5K
Gerard van Enk
Gerard van Enk@gvenk·
Al meer dan 2 weken staat deze voor onze deur, gemeente verwijst naar de supermarkt, supermarkt zegt als ze in de gelegenheid zijn en het niet te ver weg is, ze ‘m komen ophalen…ben heel benieuwd wanneer dat zal zijn. Ondertussen trekt het troep aan…
Gerard van Enk tweet media
Nederlands
17
0
8
5.7K
Google Cloud Tech
Google Cloud Tech@GoogleCloudTech·
Introducing the Open Knowledge Format (OKF), an open specification that formalizes the LLM-wiki pattern into a portable, interoperable format. AI is only as smart as the context we give it. As we build more advanced, agentic AI systems, they need accurate metadata and context to be useful. But in most organizations, that context is locked inside fragmented data catalogs, isolated wikis, scattered code comments, or the minds of senior engineers. Every time a new AI agent is built, teams are forced to solve the exact same context-assembly problem from scratch. To solve this, we've announced OKF, a vendor-neutral, open specification that formalizes the "LLM-wiki pattern" into a portable, interoperable format. It provides a standardized way to represent the enterprise knowledge that modern AI systems rely on. — Just markdown: readable in any editor, renderable on GitHub, indexable by any search tool — Just files: shippable as a tarball, hostable in any git repo, mountable on any filesystem — Just YAML frontmatter: for the small set of structured fields that need to be queryable: type, title, description, resource, tags, and timestamp We’ve also shipped reference implementations to help you hit the ground running, including an enrichment agent for BigQuery, a static HTML visualizer, and live sample bundles on @github → goo.gle/4uGvAEe ➕ Knowledge Catalog can now natively ingest OKF! Stop reinventing data models and building bespoke integrations for every new AI tool. Here's more about how OKF works → goo.gle/4uGvBbg
Google Cloud Tech tweet media
English
130
598
5K
437.4K
Dimitris Papailiopoulos
Dimitris Papailiopoulos@DimitrisPapail·
I just landed on Marseille and everybody keeps talking about a big fat cat. I don’t get it.
English
4
1
123
8.8K
Mark Pors 🦖
Mark Pors 🦖@pors·
@wesbos Yes, lots of variety, but not very tasty. It is just different ways of processing and packaging. Not real food. Same in most of the US supermarkets, I guess.
English
0
0
0
122
Wes Bos
Wes Bos@wesbos·
American abundance is nothing. This is half of the *spreadable meat* section in a Dutch grocery store
Wes Bos tweet media
@levelsio@levelsio

My biggest shock today was that America still has post offices, like ones you can actually walk into with your stuff and package it and they help you send it out We don't have them in most of Europe anymore You have tiny little counters in dirty tobacco shop and a tiny DPD or DHL or PostNL branded counter with agressive staff that really do not want to help you, and no they don't have boxes Here's a FedEx post office we walked in today, they have envelopes, boxes, bubble wraps, everything! You'll say but FedEx is private, yes you mean like most European postal companies now that are also private? Just something as basic as this is abundance and we don't have it anymore in Europe

English
25
0
101
39.2K
Mark Pors 🦖 retuiteado
Mark Pors 🦖
Mark Pors 🦖@pors·
Most neural nets are still based on the model of a neuron as proposed in the 1950's: u = activation(w·x + b) In a new paper, researchers propose a more accurate model of a biological brain neuron and found that it has quite a few advantages, like needing less training data.
Mark Pors 🦖 tweet media
English
33
123
1.3K
91.1K
hayden
hayden@haydendevs·
finally looking into openclaw and hermes agent, i feel like a boomer man, i just don't get the use case
English
418
53
3.6K
809.8K
Jared Friedman
Jared Friedman@snowmaker·
If you want to find a truly impactful idea, a good place to look is in the trash bin of history. Most of the ideas that have most broken out in the last 5 years aren't new. They're old ideas that got written off too early. examples: 1) AI: When Sam started OpenAI, saying "AGI" got you laughed at 2) Space: Elon started SpaceX after people had written off the 1960s-era dreams as sci-fi 3) Supersonic flight: Blake started Boom after everyone assumed Concorde already proved it couldn't work 4) Nuclear energy: fission and fusion are roaring back decades after ambitious people stopped studying nuclear physics 5) GLP-1's: After the fen-phen disaster, weight loss drugs were synonymous with snake oil. There's a good reason this keeps happening. When a hyped idea fails, there's a backlash. It becomes embarrassing to work on - anyone still working on it is assumed too dumb to know better. So, if you want to work on something like this, you have to teach yourself to feel the tinge of that embarrassment and push past it.
English
130
132
1.7K
178.6K

Descubrir

@IreneSolaiman @mitsuhiko @ClintFiore @scaling01 @raphaelschaad @ryolu_ @cellinlab @cjzafir