Tootfinder

Interesting, "GPT-style models have a fixed memorization capacity of approximately 3.6 bits per parameter."
https://venturebeat.com/ai/how-much-information-do-llms-really-memorize-now-we-know-thanks-to-met…

How much information do LLMs really memorize? Now we know, thanks to Meta, Google, Nvidia and Cornell
Using a clever solution, researchers find GPT-style models have a fixed memorization capacity of approximately 3.6 bits per parameter.

Update on the minirack: it looks like the Minisforum MS-A2s have cleared customs maybe. UPS isn't 100% sure what's going on from the look of the tracking information considering they have them clearing customs two months ago before they were shipped. However, I'm thinking they'll be here in a week or so and I can finally get them installed and cabled up.
#minirack

This https://arxiv.org/abs/2409.18858 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCR_…

Predicting memorization within Large Language Models fine-tuned for classification
Large Language Models have received significant attention due to their abilities to solve a wide range of complex tasks. However these models memorize a significant proportion of their training data, posing a serious threat when disclosed at inference time. To mitigate this unintended memorization, it is crucial to understand what elements are memorized and why. This area of research is largely unexplored, with most existing works providing a posteriori explanations. To address this gap, we pro…

‘We do this to survive’: Harvesting opium poppies in Myanmar’s Shan State | Drugs News | Al Jazeera
https://www.aljazeera.com/features/2025/6/7/we-do-this-to-survive-harvesting-opium-poppies-in-myanmars-shan-state?traffic_source=rss

Trade-offs in Data Memorization via Strong Data Processing Inequalities
Vitaly Feldman, Guy Kornowski, Xin Lyu
https://arxiv.org/abs/2506.01855 https://

Trade-offs in Data Memorization via Strong Data Processing Inequalities
Recent research demonstrated that training large language models involves memorization of a significant fraction of training data. Such memorization can lead to privacy violations when training on sensitive user data and thus motivates the study of data memorization's role in learning. In this work, we develop a general approach for proving lower bounds on excess data memorization, that relies on a new connection between strong data processing inequalities and data memorization. We then demonst…

“At least memorise that #tolerance becomes crime when applied to evil.”
– 6th chapter of #ThomasMann’s “Magic Mountain”
🔗

Thomas Mann – Wikiquote

Elon Musk, JD Vance, and Peter Thiel are executing a long-planned agenda to take over the government.
Donald Trump is a dementia-addled old man along for the ride.
Hopefully, the chaos makes a little more sense now.
I told you you weren't gonna like it.

Democracy Is Done: The Rise of Corporate Monarchy
The Real Agenda Behind Elon's Coup & Trump's Chaotic First Weeks

High-Speed Ultra-Energy-Efficient Memristor-Based Massive MIMO SIC Detector Circuit with Hybrid Analog-Digital Computing Architecture
Jia-Hui Bi, Shaoshi Yang, Sheng Chen, Ping Zhang
https://arxiv.org/abs/2506.03495

High-Speed Ultra-Energy-Efficient Memristor-Based Massive MIMO SIC Detector Circuit with Hybrid Analog-Digital Computing Architecture
The emerging memristor crossbar array based computing circuits exhibit computing speeds and energy efficiency far surpassing those of traditional digital processors. This type of circuits can complete high-dimensional matrix operations in an extremely short time through analog computing, making it naturally applicable to linear detection and maximum likelihood detection in massive multiple-input multiple-output (MIMO) systems. However, the challenge of employing memristor crossbar arrays to eff…

On the weak and strong Lefschetz properties for initial ideals of determinantal ideals with respect to diagonal monomial orders
Hongmiao Yu
https://arxiv.org/abs/2506.05193

On the weak and strong Lefschetz properties for initial ideals of determinantal ideals with respect to diagonal monomial orders
We study the weak and strong Lefschetz properties for $R/\mathrm{in}(I_t)$, where $I_t$ is the ideal of a polynomial ring $R$ generated by the $t$-minors of an $m\times n$ matrix of indeterminates, and $\mathrm{in}(I_t)$ denotes the initial ideal of $I_t$ with respect to a diagonal monomial order. We show that when $I_t$ is generated by maximal minors (that is, $t=\mathrm{min}\{m,n\}$), the ring $R/\mathrm{in}(I_t)$ has the strong Lefschetz property for all $m$, $n$. In contrast, for $t<\mathrm…

📉 Chart of the Week: 'Dire Picture' for BTC Miners as Revenue Flatlines Near Record Low
https://www.coindesk.com/business/2025/04/20/chart-of-the-day-dire-picture-for-btc-miners-as-revenue-flatlines-near-r…

Chart of the Week: 'Dire Picture' for BTC Miners as Revenue Flatlines Near Record Low
Despite bitcoin trading around $84,000, miner revenue has decreased due to the recent halving event and rising operational costs.