Boltzbit previews paper claiming BAST lets LLMs learn up to 1,000x faster than SOTA training
jmhernandez233 · x · 2026-09-22
Boltzbit released a preview of its paper Infinite-Parameter LLMs — Generating and Adapting Weights from Live Data, introducing Bayesian Self-learning Transformers (BAST).
- The team claims LLMs built on BAST can learn up to 1,000x faster than human-engineered SOTA training algorithms, framing it as a shift from cost-heavy static-weight AI to energy-efficient dynamic-weight AI
- Motivation: pretraining text is projected to run out within 2-5 years, while AI agents generate vast amounts of continuously growing session data that is currently lost once sessions end
- The core idea is generating and adapting model weights directly from live data for continual learning
Caveat: this is a self-published preview; the 1,000x figure is the team's own claim with no third-party replication or peer review yet.
More from Research
- Anthropic Is Setting Up a Biology Lab Where Claude Guides Robots Through Drug Experiments — The Decoder · 2026-09-22
- GAE: A Geometry-Native Autoencoder Cuts World Model FVD by 23.1% — yshan2u · 2026-09-22
- Over 1TB of China A-share Level-2 limit order book tick data hits Hugging Face — venvoo · 2026-09-22
- Reddit proposes measuring LLMs by cost per accepted task, not cost per token, after Grok 4.7 launch — Crescitaly · 2026-09-22
- Google's ScientistTwo solves 80.4% of 107 top-venue ML problems autonomously — thisdudelikesAI · 2026-09-22
- Tencent Hunyuan's WebCraftBench tests web apps like software, matching human preference 85.3% — TencentHunyuan · 2026-09-22