Halcyon

Headlines · 2

1. Oct 21
   US Utilities Tighten Grid Rules for AI Data Centers

   US utilities and grid operators are demanding higher rates, longer contracts, collateral and steep engineering study fees from data-center developers to screen out speculative or duplicate grid requests. TechRadar, citing Wood Mackenzie and other reporting, says US developers have requested 1,066GW of connections and that utilities are expected to commit to about 28%. After Exelon introduced transmission security agreements, its high-probability data-center load fell from 18GW at the end of 2025 to about 11GW in the second quarter. The report does not establish how much of the requested capacity would serve AI workloads rather than other data-center uses.

   TechRadar · 🔥 0
2. Oct 18
   Halcyon paper claims strong text-to-image generation without language pretraining

   An arXiv paper posted October 19, "Text-to-Image Generative Modeling Without Language Pretraining," by Keshigeyan Chandrasegaran, Li Fei-Fei, Michael Poli and colleagues, asks whether language pretraining is necessary for strong text-to-image generation. The authors introduce Halcyon, a family of early-fusion multimodal autoregressive models pretrained entirely from scratch on GPIC, a dataset of 100 million image-text pairs. They report that Halcyon-0.52B beats LlamaGen-XL/8 on gFD (48.0 vs. 63.3) with 18.2 times less pretraining compute, and that Halcyon-1.6B beats JiT-T2I-XXL/8++ (35.3 vs. 41.0) with 275 times less compute; controlled experiments found text-only pretraining did not improve T2I generation, and code and models are released.

   Hugging Face · Papers · 🔥 0

Experience and discussion from the community

Share my Halcyon experienceAsk about Halcyon