/
Navigation
C
Chronicles
Browse all articles
C
E
Explore
Semantic exploration
E
R
Research
Entity momentum
R
N
Nexus
Correlations & relationships
N
~
Story Arc
Topic evolution
S
Drift Map
Semantic trajectory animation
D
P
Posts
Analysis & commentary
P
Browse
@
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
?
Concept Search
Semantic similarity search
!
High Impact Stories
Top coverage by position
+
Sentiment Analysis
Positive/negative coverage
*
Anomaly Detection
Unusual coverage patterns
Analysis
vs
Rivalry Report
Compare two entities head-to-head
/\
Semantic Pivots
Narrative discontinuities
!!
Crisis Response
Event recovery patterns
Connected
Nav: C E R N
Search: /
Command: ⌘K
Embeddings: large
VOICE ARCHIVE

Lucas Beyer

@giffmana
28 posts
2026-02-25
Holy cow! I hope for the sake of my friends there, that it won't happen... but Anthro -of all the labs!- being the one forced to train a WarClaude would be the most ironic outcome possible! Btw, completely unrelated, but i noticed that if you set the learning rate just 3x too [image]
2026-02-25 View on X
Reuters

Source: Anthropic has no intention of easing Claude usage restrictions for military purposes, following Dario Amodei's meeting with Pete Hegseth

Artificial intelligence lab Anthropic has no intention of easing its usage restrictions for military purposes, a person familiar with the matter …

Holy cow! I hope for the sake of my friends there, that it won't happen... but Anthro -of all the labs!- being the one forced to train a WarClaude would be the most ironic outcome possible! Btw, completely unrelated, but i noticed that if you set the learning rate just 3x too [image]
2026-02-25 View on X
Axios

Sources: DOD told Anthropic it will invoke the Defense Production Act or label Anthropic a “supply chain risk”, if not given unfettered Claude access by Friday

Defense Secretary Pete Hegseth gave Anthropic CEO Dario Amodei until Friday evening to give the military unfettered access …

2026-02-12
Couple interesting things (new to me): - Macrohard is, roughly, computer-use agents. - xAI wants to lean much more into coding model. - Elon would find it epic to build a satellite factory and catapult ("mass driver") on the moon lol
2026-02-12 View on X
The Information

xAI all-hands: Nikita Bier says X hit $1B in ARR from selling $3- to $40-per-month subscriptions, but offered no details on ad revenue or other xAI financials

and both are smarter than expected. 😅Logan Jastremski /@loganjastremski:My main takeaway is that engineers need to run the show and someone needs a large hammer to unblock teams an...

Couple interesting things (new to me): - Macrohard is, roughly, computer-use agents. - xAI wants to lean much more into coding model. - Elon would find it epic to build a satellite factory and catapult ("mass driver") on the moon lol
2026-02-12 View on X
CNBC

Elon Musk announces xAI implemented a reorganization that “required parting ways with some people”, after two xAI co-founders said they were leaving this week

SpaceX CEO Elon Musk on Wednesday announced the the company's xAI artificial intelligence venture implemented a reorganization that …

2025-12-13
Not sure if it's Tinker's SFT or Qwen3-VL-235B that's not good at this (and dino2 even worse?), but guys, ... fine-grained few-shot SFT we already did better than this in 2020 with good old BiT, which was 928M. (Colleagues made an internal “as-a-service” eventually too, lol!)
2025-12-13 View on X
Thinking Machines Lab

Mira Murati's Thinking Machines Lab makes Tinker, its API for fine-tuning language models, generally available, adds support for Kimi K2 Thinking, and more

Tinker is a dream for multi-agent setups, Nathan Lambert / @natolambert : Please add olmo3 @johnschulman2 et al. The goal is to make it the foundational research infrastructure for...

2025-11-26
Sorry, my little one wrote this!
2025-11-26 View on X
CNBC

Nvidia says it is “delighted by Google's success” but “Nvidia is a generation ahead”, after a report that Meta could use Google's TPUs in its data centers

Nvidia on Tuesday said its tech remains a generation ahead of the industry, in response to Wall Street's concerns … Source: @nvidianewsroom .

2025-11-25
Yoooo looks like Denny deleted his “game over” tweet on Big Opus Day. Probably nothing though lol [image]
2025-11-25 View on X
Anthropic

Anthropic launches Claude Opus 4.5, saying it is “the best model in the world for coding, agents, and computer use” and “meaningfully better at everyday tasks”

Our newest model, Claude Opus 4.5, is available today.  It's intelligent, efficient …

2025-07-01
They have a cloud infrastructure?
2025-07-01 View on X
Bloomberg

Sources: Mike Rockwell and Craig Federighi started a project to evaluate external models to power Siri; after testing, Anthropic's tech seemed most promising

Apple Inc. is considering using artificial intelligence technology from Anthropic PBC or OpenAI to power a new version of Siri …

2025-04-06
10M context length for llama4! This is the moment I've been waiting for to plug one of my fun projects: https://longcat.wtf/ Sound ON! (And congrats to the team who made the 10M contest happen. Gemini is not alone anymore) [image]
2025-04-06 View on X
Meta

Meta launches Llama 4 Maverick with 400B parameters and Scout with 109B parameters and a 10M context window, and previews Behemoth with 2T total parameters

Takeaways  — We're sharing the first models in the Llama 4 herd, which will enable people to build more personalized multimodal experiences.

2025-02-22
The SakanaAI story isn't over yet... If the inputs to tests are constants, then yeah returning the expected constant is one way to get massive speedup!
2025-02-22 View on X
TechCrunch

Sakana AI walks back claims that its new AI CUDA Engineer can speed up AI training by up to 100x, after complaints about worse-than-average training performance

Kyle Wiggers / TechCrunch :

2024-12-28
This actually reproduces as of today. In 5 out of 8 generations, DeepSeekV3 claims to be ChatGPT (v4), while claiming to be DeepSeekV3 only 3 times. Gives you a rough idea of some of their training data distribution. [image]
2024-12-28 View on X
TechCrunch

DeepSeek-V3 sometimes identifies itself as ChatGPT when asked which model it is; some speculate that its training datasets may contain text generated by ChatGPT

Kyle Wiggers / TechCrunch :

It's not too hard to probe a model and deduce how the various parts of its training were done. Once you know that, making it behave a certain way (eg bypass its safety tuning) can be relatively easily: [image]
2024-12-28 View on X
TechCrunch

DeepSeek-V3 sometimes identifies itself as ChatGPT when asked which model it is; some speculate that its training datasets may contain text generated by ChatGPT

Kyle Wiggers / TechCrunch :

2024-10-26
Waymo is really taking off now (in the US), this is great! And slowly but surely moving to more seasonal weather zones, as opposed to the current always-sunny ones. Kudos to the team. [image]
2024-10-26 View on X
CNBC

Waymo closed a $5.6B Series C led by Alphabet to expand its robotaxi service in and beyond LA, San Francisco, and Phoenix, bringing its total funding to $11.1B

Having had the pleasure of seeing Waymo come … Lora Kolodny : Waymo has closed its series C funding round at $5.6 billion.  The self-driving venture owned by Google parent Alphabet...

2024-08-15
Guess the gang and i are doing something wrong then...
2024-08-15 View on X
Wall Street Journal

Eric Schmidt says he “misspoke” when he said Google was losing the AI race because the company decided that “working from home was more important than winning”

Ex-Google CEO had said the tech company's work-life balance was ‘more important than winning’

2024-06-13
If you've never been at BigCo, you've likely found yourself thinking something like: > why the fuck is this BigCo thing so obviously shitty? Fixing this would take me one afternoon! Their engineers/designers must be trash. The screenshot below perfectly describes why: [image]
2024-06-13 View on X
Fortune

Sources detail Amazon's struggles to build a new generative AI-powered Alexa, including privacy concerns keeping Alexa's teams from using Anthropic's Claude

“Alexa, let's chat.”  —  With that phrase, David Limp, at the time Amazon's head of devices and services, showed off …

2024-03-13
I read all of this, and my brain just goes: “I really hope they are letting Hugo train some yolo runs on it without too much process headache, given his track record!”
2024-03-13 View on X
DatacenterDynamics

Meta details its two new data center scale clusters, both containing 24,576 Nvidia H100 GPUs that the company is using for AI workloads like training Llama 3

Charlotte Trueman / DatacenterDynamics :

2024-02-14
Friends telling me his last two yolo-runs were only given 256 and 64 GPUs, respectively, so he was like “fuck that shit”.
2024-02-14 View on X
The Information

AI researcher Andrej Karpathy, one of OpenAI's founding members, leaves the startup; Karpathy was developing a product that he had described as an AI assistant

Andrej Karpathy, one of the founding members of OpenAI, has left the company, a spokesperson confirmed.

2023-11-20
TL;DR: researcher politics start really ramping up the moment compute gets scarce.
2023-11-20 View on X
The Atlantic

Sources describe how ChatGPT's success widened OpenAI's ideological rifts, as Ilya Sutskever began behaving like a spiritual leader anticipating AGI's arrival

Sam Altman's weekend of shock and drama began a year ago, with the release of ChatGPT.  —  To truly understand the events …

TL;DR: researcher politics start really ramping up the moment compute gets scarce.
2023-11-20 View on X
TechCrunch

Satya Nadella says Sam Altman, Greg Brockman, and OpenAI staff will join Microsoft's new “advanced AI research team” and Microsoft remains committed to OpenAI

Microsoft has hired OpenAI co-founders Sam Altman and Greg Brockman to head up a “new advanced AI research team,” …

2023-11-17
This model is reeeeeeally good
2023-11-17 View on X
Wired

YouTube previews Dream Track and Music AI tools, which use DeepMind's new model Lyria to generate music in the style of famous artists, from humming, and more

Will Knight / Wired :