Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 118

Delivering 1.5 M TPS Inference on NVIDIA GB200 NVL72, NVIDIA Accelerates OpenAI gpt-oss Models from Cloud to Edge

NVIDIA and OpenAI began pushing the boundaries of AI with the launch of NVIDIA DGX back in 2016. The collaborative AI innovation continues with the OpenAI gpt-oss-20b and gpt-oss-120b launch. NVIDIA has optimized both new open-weight models for accelerated inference performance on NVIDIA Blackwell architecture, delivering up to 1.5 million tokens per second (TPS) on an NVIDIA GB200 NVL72 system.

The gpt-oss models are text-reasoning LLMs with chain-of-thought and tool-calling capabilities using the popular mixture of experts (MoE) architecture with SwigGLU activations. The attention layers use RoPE with 128k context, alternating between full context and a sliding 128-token window. The models are released in FP4 precision, which fits on a single 80 GB data center GPU and is natively supported by Blackwell.

The models were trained on NVIDIA H100 Tensor Core GPUs, with gpt-oss-120b requiring over 2.1 million hours and gpt-oss-20b about 10x less. NVIDIA worked with several top open-source frameworks such as Hugging Face Transformers, Ollama, and vLLM, in addition to NVIDIA TensorRT-LLM for optimized kernels and model enhancements. This blog post showcases how NVIDIA has integrated gpt-oss across the software platform to meet developers’ needs.

Tesla’s New Strategy Has Uber Terrified

Questions to inspire discussion.

👥 Q: What is the ratio of robo taxis to supervisors in Tesla’s network? A: Tesla’s robo taxi network operates with a 10:1 ratio of robo taxis to supervisors, enabling efficient management and cost-effective operations.

Market Disruption.

📊 Q: How is Whim, a Tesla competitor, performing in the market? A: As of April 2025, Whim has 25% of San Francisco gross bookings, surpassing Lyft, with an average price of $20 per mile compared to Uber’s $15 and Lyft’s $14.

Technology Superiority.

🤖 Q: How does Tesla’s robo taxi software compare to human drivers? A: Tesla’s robo taxi software has crossed the uncanny valley, providing a smooth and comfortable driving experience similar to a human chauffeur, outperforming Uber’s inconsistent service.

Continue reading “Tesla’s New Strategy Has Uber Terrified” | >

“They’re Handing Supercomputer Power to Everyone”: Nexus Ignites Fierce Debate as Georgia Tech’s 400 Quadrillion-Operation AI Revolution Promises Unmatched Access for US Scientists

IN A NUTSHELL 🚀 Nexus aims to democratize access to advanced computing with its groundbreaking capabilities. 🔬 The supercomputer is set to revolutionize scientific research with over 400 quadrillion operations per second. 🌍 Nexus’s accessibility philosophy ensures researchers nationwide can utilize its powerful AI tools. 🔗 Georgia Tech collaborates with the University of Illinois for

First 3D-Printed Home Made Primarily From Soil is Built in Japan–Ditching Unsustainable Concrete

Collaborating with robotics engineers and Italian 3D printer manufacturers, a Japanese company is building “homes of earth” made primarily from soil.

Utilizing AI technology from design through construction, Lib Work, Ltd. completed their first 3D-printed earth home in Yamaga, Kumamoto on July 22, calling their creative process “uncharted territory where tradition and convention offered no guide”

While the automotive industry has undergone rapid transformation through technological advances, the housing industry has seen virtually no fundamental innovation in construction methods, materials, or structures for over 50 years.

Neural Networks Go Nano: Brain-Inspired Learning Takes Flight

For the first time, a physical neural network has successfully been shown to learn and remember ‘on the fly’, in a way inspired by and similar to how the brain’s neurons work.

The result opens a pathway for developing efficient and low-energy machine intelligence for more complex, real-world learning and memory tasks.

Published today (November 1) in Nature Communications, the research is a collaboration between scientists at the University of Sydney and the University of California at Los Angeles (UCLA).

A Scalable Artificial Neuron Based on Ultrathin Two-Dimensional Titanium OxideClick to copy article linkArticle link copied!

A spiking neural network consists of artificial synapses and neurons and may realize human-level intelligence. Unlike the widely reported artificial synapses, the fabrication of large-scale artificial neurons with good performance is still challenging due to the lack of a suitable material system and integration method. Here, we report an ultrathin (less than10 nm) and inch-size two-dimensional (2D) oxide-based artificial neuron system produced by a controllable assembly of solution-processed 2D monolayer TiOx nanosheets. Artificial neuron devices based on such 2D TiOx films show a high on/off ratio of 109 and a volatile resistance switching phenomenon. The devices can not only emulate the leaky integrate-and-fire activity but also self-recover without additional circuits for sensing and reset. Moreover, the artificial neuron arrays are fabricated and exhibited good uniformity, indicating their large-area integration potential. Our results offer a strategy for fabricating large-scale and ultrathin 2D material-based artificial neurons and 2D spiking neural networks.

Artificial neurons based on antiferromagnetic auto-oscillators as a platform for neuromorphic computing

A spiking neural network consists of artificial synapses and neurons and may realize human-level intelligence. Unlike the widely reported artificial synapses, the fabrication of large-scale artificial neurons with good performance is still challenging due to the lack of a suitable material system and integration method. Here, we report an ultrathin (less than10 nm) and inch-size two-dimensional (2D) oxide-based artificial neuron system produced by a controllable assembly of solution-processed 2D monolayer TiOx nanosheets. Artificial neuron devices based on such 2D TiOx films show a high on/off ratio of 109 and a volatile resistance switching phenomenon. The devices can not only emulate the leaky integrate-and-fire activity but also self-recover without additional circuits for sensing and reset. Moreover, the artificial neuron arrays are fabricated and exhibited good uniformity, indicating their large-area integration potential. Our results offer a strategy for fabricating large-scale and ultrathin 2D material-based artificial neurons and 2D spiking neural networks.

Small but mighty: A seed-inspired monocopter idea takes flight

From a seed-inspired design to a 26-minute flight time on a single rotor, a new monocopter developed by SUTD researchers marks a 10-year journey towards redefining how efficient small flying robots can be.

When Singapore celebrated its 50th year of independence in 2015, a team of student researchers led by Associate Professor Foong Shaohui from Singapore University Technology and Design (SUTD) embarked on an ambitious challenge: to design and build a drone capable of 50 minutes of sustained flight.

At the time, most hobbyist quadcopters could barely manage half of that. The SG50 Multi-Rotor Drone project succeeded, but to fly that long, the craft had to be large, complex, and heavy.

/* */