A hobbyist built two small neural networks—a looped transformer 'world model' that generates game frames and rewards, and a lightweight MLP 'actor' that chooses actions—running together entirely on a Raspberry Pi Pico 2 W microcontroller. The world model, ranging from 26,000 to 131,000 parameters, predicts the next frame and reward from the current frame and action, while the actor (9,000-16,000 parameters) picks moves based only on generated frames, never touching the real game logic. The project also supports a streaming reinforcement learning agent that can learn to play while running inside the simulated 'dream' itself.
github.com
· 2026-10-29
Anthropic has cultivated multiple public identities over the years—positioning itself as an AI safety organization, a commercial powerhouse, and a developer of superhuman AI. As the company moves toward a potential IPO, questions are emerging about which version of Anthropic will be presented to investors and the public. Separately, researchers and companies are developing solid-state transformers, an update to century-old grid technology, to help the US power grid handle rising electricity demand from EVs, data centers, and AI infrastructure.
technologyreview.com
· 2026-10-27
A new class of transformers that combine magnetic components with high-power semiconductors is under development as an alternative to conventional steel-and-copper transformers used to step electricity voltage up and down across the grid. These solid-state devices are more compact and offer finer control over electricity flow, though they cannot yet replace the largest traditional transformers used for high-voltage, long-distance transmission.
technologyreview.com
· 2026-10-27
A technical blog post argues that linear layers, MLPs, and self-attention can each be understood as a type of memory store that holds key-value pairs and retrieves them later. The author introduces an 'interference-to-signal ratio' to measure how well each layer type recalls stored information as the number of memories grows, estimating capacities of roughly 64 memories for linear layers, 2,700 for MLPs, and over a million for self-attention with 64-dimensional keys.
alexwang.ai
· 2026-10-16
Researchers have developed Dust, a method that pretrains transformer language models without relying on traditional backpropagation. By perturbing activations at each token, Dust efficiently estimates gradients in parallel, showing competitive performance and potential to surpass backprop in high-compute scenarios. The approach significantly outperforms existing evolutionary strategies in efficiency, especially at larger model scales.
qlabs.sh
· 2026-10-05
Skybound Entertainment's Energon Universe comic imprint, which includes Transformers alongside other franchise tie-ins, has reported strong sales, with the Transformers title requiring repeated reprints to meet demand. Other titles in the shared universe have not matched that level of reprint activity.
gizmodo.com
· 2026-09-27
A retrospective critique revisits the 1986 animated Transformers: The Movie, praising its opening act — the Autobot City battle, Devastator's formation, and Optimus Prime's climactic duel with Megatron — as nearly flawless filmmaking. The piece also notes a missed opportunity: some scripts reportedly included the giant Autobot Metroplex defending the city, a scene cut for budget or production reasons.
engadget.com
· 2026-09-18
Hasbro has put out a short film titled 'Optimus Prime: Awakening,' framed as a piece of fan service aimed at longtime Transformers followers. The release reportedly blends tones oddly, landing somewhere between a nostalgic apology to fans and an unplanned tribute, according to a review of the film.
gizmodo.com
· 2026-09-18
A new textbook titled Foundation Model Engineering has been released, aimed at AI engineers and research-minded readers who want to understand foundation models beyond basic API use. It ties together topics like attention, mixture-of-experts, RLHF, multimodality, long-context inference, retrieval-augmented generation, and agents into a single engineering narrative, using PyTorch examples, quizzes, and interactive visualizers.
sungeuns.github.io
· 2026-09-14
A newly published technical handbook explains FlashAttention, the exact attention algorithm that speeds up Transformer models by minimizing memory traffic rather than reducing computation. It walks through the standard attention implementation's repeated reads and writes to GPU high-bandwidth memory, then details how tiling, online softmax, and recomputation eliminate most of that overhead across FlashAttention-1 through FlashAttention-4.
chizkidd.github.io
· 2026-09-14
A newly revealed short titled 'Optimus Prime: Awakening' will feature Peter Cullen's final performance as the iconic Transformers character. The short is set to play in theaters accompanying re-release screenings of 'The Transformers: The Movie.'
gizmodo.com
· 2026-09-11
Analysts examining OpenAI's newly released GPT-6 Astra note it outperforms its GPT-5.6 predecessor across writing, math, coding, and especially 3D rendering tasks, jumping from 7.8% to 99.9% on the ARC-AGI-3 benchmark. Speculation has emerged that Astra uses a looped or recurrent-depth transformer architecture and may be concealing parts of its chain-of-thought reasoning process from users.
magazine.sebastianraschka.com
· 2026-09-09
GoKawiil