CoreweaveAppleAI EngineerNVIDIAGoogleMetaLambdaUber+4

CoreweaveBlog1w

ART-Optimized Megatron (AOM): 12x higher training throughput

AppleBlog1w

SCLATE: A Substrate for Continual-Learning Agent Training and Evaluation

AI EngineerYouTube2w

Teaching LLMs to Speak Spotify — Yves Raimond & Jacqueline Wood, Spotify

NVIDIABlog2w

Efficient MoE Training for Biological Foundation Models

GoogleBlog2w

Reproducing OLMo 3 7B Pre-training in MaxText: case study of large scale training on TPUs- Google Developers Blog

MetaBlog3 Aug

GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model

NVIDIABlog10 Jul

Reducing High-Bandwidth Memory Bottlenecks in JAX-Based LLM Training with Host Offloading

NVIDIABlog6 Jul

Enhancing Goodput in Large-Scale LLM Training with Nonuniform Tensor Parallelism

LambdaBlog16 Jun

MLPerf Training v6.0: Lambda delivers fastest LLM training on NVIDIA GB300 NVL72 and fastest MoE training on NVIDIA HGX B200

NVIDIABlog8 Jun

Train Models Faster with JAX and MaxText Using NVFP4 on NVIDIA Blackwell

CoreweaveBlog20 May

CoreweaveBlog20 May

MLPerf Results: A Record-Breaking, Cloud-Native AI Supercomputer

CoreweaveBlog20 May

How the MosaicML Platform Makes Large-Scale Model Training More Accessible

CoreweaveBlog20 May

NovelAI as Among the First to Have NVIDIA HGX H100 GPUs Online

CoreweaveBlog20 May

AI storage and LLMs: 4 critical needs to look out for

UberBlog14 Apr

Open Source and In-House: How Uber Optimizes LLM Training

BasetenBlog31 Mar

Open-source LLM training is a mess. Here is how it all works.

MetaBlog29 Sept

Meta’s Infrastructure Evolution and the Advent of AI

Hugging FaceBlog26 Jul

Ulysses Sequence Parallelism: Training with Million-Token Contexts

CerebrasBlog17 Jul

Introducing gigaGPT: GPT-3 sized models in 565 lines of code

Showing the newest 40 of 42 posts.