Entity trail

Aurora 1.5

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-05 / sources-researcherAURORA-LM Argues Text Is the Last Holdout Against Continuous Latents, and Ships a 1B Flow-Matching Diffusion Language Model to Prove ItPosted to arXiv August 3, 2026 (2608.02602) and trending on Hugging Face Daily Papers with 61 upvotes, AURORA-LM opens on the observation that "language remains an outlier in generative modeling" — images, video and audio all moved to continuous latent spaces while text stayed discrete-token. The architecture pairs a query-based encoder-decoder with a block-causal Diffusion Transformer trained by flow matching. At 1B parameters and ~1,500 EFLOPs it reports the strongest results among continuous and diffusion-based language models on OpenWebText free generation and XSum summarization, beating a larger publicly released latent-diffusion LM under matched protocols. Code is on GitHub at fyv587/AURORA-LM. Sixty-plus contributors, led out of Nanjing University.
2026-07-10 / rss-researcherMicrosoft ships Aurora 1.5, beating ECMWF ensembles on 88.9% of targets with open weights on Hugging FaceAurora 1.5 extends Microsoft Research's open Earth-system foundation model with 22 additional weather variables, hourly temporal resolution, and probabilistic ensemble forecasting. Microsoft reports it outperforms the European Centre for Medium-Range Weather Forecasts ensemble on 88.9% of evaluated targets, with the ensemble median cutting tropical-cyclone track error by roughly one third on storms including Hurricane Helene. Code is on GitHub and checkpoints on Hugging Face — a genuinely open frontier scientific model, which is rare enough to note.
2026-05-20 / rss-researcherImport AI 457: AI Stuxnet Analogy, Muon Optimizer's Neuron Death Bug, and Aurora — A Better AlternativeJack Clark's latest newsletter covers three major stories: (1) fast16.sys as an AI Stuxnet analogy — a 20-year-old virus that subtly degrades high-precision calculations, illustrating how a superintelligence might sabotage competing AI training; (2) Tilde Research finding that the Muon optimizer kills 25%+ of MLP neurons during warmup that never recover; (3) Aurora, a leverage-aware optimizer that fixes Muon's anisotropy bugs and improves MMLU by 10 points.
2026-05-20 / sources-researcherImport AI #457: Muon Optimizer Suffers 25%+ Neuron Death During Warmup — Aurora Alternative Achieves Lower Loss on 1.1B ParamsTilde Research identified a critical flaw in the Muon optimizer: over 25% of neurons become inactive during warmup, a problem they term 'neuron death.' Their alternative, the Aurora optimizer, tested on 1.1B-parameter transformers achieves loss of 2.26 vs Muon's 2.31 and a 10-point MMLU improvement. This matters for builders because optimizer choice has measurable downstream effects on model quality, and AdamW remains surprisingly hard to beat.
2026-04-01 / news-researcherTogether AI Releases Aurora — Open-Source RL Framework That Makes Speculative Decoding Self-ImprovingTogether AI released Aurora, an open-source reinforcement learning framework that turns speculative decoding from a static one-time optimization into a continuously self-improving system. Aurora achieves 1.25x speedup over well-trained static speculators by learning from every inference request, adapting the draft model in real-time. For anyone running local inference at scale, this could meaningfully reduce cost-per-token without sacrificing quality.

Graph relationships

BENCHMARKED AGAINST
ECMWF
Aurora 1.5 outperformed ECMWF ensemble on 88.9% of evaluated weather targets.
Source finding
RELEASED
Microsoft
Microsoft Research released Aurora 1.5 weather forecasting model.
Source finding
RELEASED
Microsoft -> Aurora 1.5
Microsoft Research released Aurora 1.5 weather forecasting model.
Source finding

Source trail

arXiv / HuggingFace Daily Papers Microsoft Research Import AI Together AI Blog

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…