Researchers baked 3x inference speedups directly into LLM weights — without speculative decoding

Researchers from the University of Maryland, Lawrence Livermore, Columbia and TogetherAI have developed a training technique that triples LLM inference speed without auxiliary models or infrastructure ...
InfoWorld

Multi-token prediction technique triples LLM inference speed without auxiliary draft models

With reported 3x speed gains and limited degradation in output quality, the method targets one of the biggest pain points in production AI systems: latency at scale.
NextBigFuture

Defeating Nondeterminism in LLM Inference by Thinking Machines

A research article by Horace He and the Thinking Machines Lab (X-OpenAI CTO Mira Murati founded) addresses a long-standing issue in large language models (LLMs). Even with greedy decoding bu setting ...
Forbes

Cerebras Now The Fastest LLM Inference Processor; Its Not Even Close

The company tackled inferencing the Llama-3.1 405B foundation model and just crushed it. And for the crowds at SC24 this week in Atlanta, the company also announced it is 700 times faster than ...
Semiconductor Engineering

LLM Inference On CPUs (Intel)

“Large language models (LLMs) have demonstrated remarkable performance and tremendous potential across a wide range of tasks. However, deploying these models has been challenging due to the ...
Semiconductor Engineering

The On-Device LLM Revolution

Users running a quantized 7B model on a laptop expect 40+ tokens per second. A 30B MoE model on a high-end mobile device ...
NextBigFuture

OpenAI Strawberry LLM Reasoning Needs More Compute and Energy for Inference

Jim Fan is one of Nvidia’s senior AI researchers. The shift could be about many orders of magnitude more compute and energy needed for inference that can handle the improved reasoning in the OpenAI ...
InfoQ

GPULlama3.java Brings GPU-Accelerated LLM Inference to Pure Java

A monthly overview of things you need to know as an architect or aspiring architect. Unlock the full InfoQ experience by logging in! Stay updated with your favorite authors and topics, engage with ...

Nvidia’s new technique cuts LLM reasoning costs by 8x without losing accuracy

Nvidia researchers developed dynamic memory sparsification (DMS), a technique that compresses the KV cache in large language models by up to 8x while maintaining reasoning accuracy — and it can be ...
Business Wire

Enfabrica Unveils Industry’s First Ethernet-Based AI Memory Fabric System for Efficient Superscaling of LLM Inference

MOUNTAIN VIEW, Calif.--(BUSINESS WIRE)--Enfabrica Corporation, an industry leader in high-performance networking silicon for artificial intelligence (AI) and accelerated computing, today announced the ...
XDA Developers on MSN

I served a 200 billion parameter LLM from a Lenovo workstation the size of a Mac Mini

This mini PC is small and ridiculously powerful.