media Hugging Face Forums · 1h ago · open_models

Patience Strong proposes building neural networks as semiconductor devices

from English

Patience Strong argues that current neural networks consume excessive compute power by emulating circuits in software rather than implementing them as physical hardware. The paper suggests replacing this emulation with a semiconductor device where weights and biases are stored in non-volatile memory.

Weights and biases are realized as Digital to Analog Converters (DACs) with non-volatile registers that survive power loss.
Activation functions are implemented using fast-switching transistors that trigger based on voltage or current thresholds.
The device processes neurons in parallel, offering potential speed improvements over GPUs and significantly lower power consumption.
Successive AI releases can be overwritten directly onto the non-volatile chip, maintaining parity with current upgrade practices.

The author contends that this approach eliminates the need for conventional computers to process the network logic, requiring them only for input and output handling.

Importance 1/3 Trust 1/3 Hugging Face Forums Hardware & chips Research paper

Read original

lab NVIDIA Research · 22d ago · 6 views

Nvidia presents 2M algorithm reducing complex matrix multiplication to two real GEMMs

A new paper introduces the 2M multiplication algorithm, which reduces the cost of complex matrix multiplication from three real General Matrix Multiplications (GEMMs) to just two. This approach applies to matrices with integer real and imaginary parts by utilizing quadratic time pre- and post-processing steps.

arxiv arXiv cs.AI · 43d ago

HAMON: Passive Optical Forecasting for Long-Horizon Time-Series

HAMON uses passive optical components to perform long-horizon time-series forecasting, outperforming top digital models on ETTm2 across all horizons and on ETTh2 at all but the longest horizon. It achieves up to 14% lower MSE and relies on physical optical propagation without trainable digital layers, demonstrating that passive optical mixing can produce competitive forecasts.

arxiv arXiv cs.LG · 43d ago

HAMON: Passive Optical Forecasting Core

HAMON uses passive optical diffraction to generate forecasts, outperforming digital baselines on ETTm2 at all horizons and ETTh2 at all but the longest horizon. It achieves up to 14% lower MSE and operates without trainable digital mixing, relying instead on physical optical propagation.

lab Hugging Face Blog · 4h ago

Ai2 releases OlmoEarth Platform for planetary-scale geospatial inference

Ai2 has launched the OlmoEarth Platform, an infrastructure system designed to take Earth observation foundation models from fine-tuning through to large-scale inference. The platform addresses the engineering challenges of processing terabytes of multimodal satellite data across multiple providers and resolutions.

arxiv arXiv cs.LG · 13h ago

Moonshot AI releases Kimi K3, a 2.8T parameter MoE model with million-token context

Moonshot AI has introduced Kimi K3, an open-source Mixture-of-Experts model featuring 2.8 trillion total parameters and 104 billion activated parameters per token. The model supports native vision capabilities and a 1-million-token context window, leveraging Kimi Delta Attention and Stable LatentMoE to achieve approximately 2.5x scaling efficiency over its predecessor, Kimi K2.