A minimal tensor processing unit (TPU), inspired by Google’s TPU

a-minimal-tensor-processing-unit-(tpu),-inspired-by-google’s-tpu

A minimal tensor processing unit (TPU), reinvented from Google’s TPU V2 and V1. tinytpu.mp4 Table of Contents Architecture Processing Element (PE) Function: Performs a multiply-accumulate operation every clock cycle Data Flow: Incoming data is multiplied by a stored weight and added to an incoming partial sum to produce an output sum Incoming data also passes […]

GenAI FOMO has spurred businesses to light nearly $40B on fire

genai-fomo-has-spurred-businesses-to-light-nearly-$40b-on-fire

US companies have invested between $35 and $40 billion in Generative AI initiatives and, so far, have almost nothing to show for it. According to a report [PDF] from MIT’s NANDA (Networked Agents and Decentralized AI) initiative, 95 percent of enterprise organizations have gotten zero return from their AI efforts. Only 5 percent of organizations […]

T-Mobile claimed selling location data without consent is legal–judges disagree

t-mobile-claimed-selling-location-data-without-consent-is-legal–judges-disagree

Carriers claimed selling data didn’t violate law Instead of denying the allegations, the carriers argued that the FCC overstepped its authority. But the appeals court panel decided that the FCC acted properly: Sprint and T-Mobile (collectively, “the Carriers”) now petition for our review. Neither denies what happened. Instead, they argue that the undisputed facts do […]

How much do electric car batteries degrade?

how-much-do-electric-car-batteries-degrade?

It’s always the battery in my mobile phone that gives up on me first. After just a few years, it can barely make it through the day without getting another charge. Most electric cars have the same types of batteries — usually lithium-ion — so the assumption is that they degrade just as quickly. This […]

TREAD: Token Routing for Efficient Architecture-Agnostic Diffusion Training

tread:-token-routing-for-efficient-architecture-agnostic-diffusion-training

Abstract:Diffusion models have emerged as the mainstream approach for visual generation. However, these models typically suffer from sample inefficiency and high training costs. Consequently, methods for efficient finetuning, inference and personalization were quickly adopted by the community. However, training these models in the first place remains very costly. While several recent approaches – including masking, […]

Left to Right Programming: Programs Should Be Valid as They Are Typed

left-to-right-programming:-programs-should-be-valid-as-they-are-typed

2025-08-17 Programs Should Be Valid as They Are Typed I don’t like Python’s list comprehensions: text = “apple banana cherryndog emu fox” words_on_lines = [line.split() for line in text.splitlines()] Don’t get me wrong, declarative programming is good. However, this syntax has poor ergonomics. Your editor can’t help you out as you write it. To see […]

Anna’s Archive: An Update from the Team

annas-archive.li/blog, 2025-08-17 About recent events. We are still alive and kicking. In recent weeks we’ve seen increased attacks on our mission. We are taking steps to harden our infrastructure and operational security. The work of securing humanity’s legacy is worth fighting for. Since we started in 2022, we have liberated tens of millions of books, […]

Who Invented Backpropagation?

who-invented-backpropagation?

Efficient backpropagation (BP) is central to the ongoing Neural Network (NN) ReNNaissance and “Deep Learning.” Who invented it? BP’s modern version (also called the reverse mode of automatic differentiation) was first published in 1970 by Finnish master student Seppo Linnainmaa [BP1] [R7]. In 2020, we celebrated BP’s half-century anniversary! A precursor of BP was published […]