Schmidhuber: long-context autoregression was cool long before 2016, cites pioneer-credit survey
SchmidhuberAI · x · 2026-09-09
Jürgen Schmidhuber pushed back on the claim that "autoregression with long context" is a recent idea, arguing it predates 2016. He also pointed to his updated survey (arXiv:2212.11279), framing machine learning as "the science of credit assignment" and crediting the pioneers of deep learning as a supplement to his award-winning 2015 survey.
More from Models
- Anthropic admits internal models used in math proof, sparking transparency row — teortaxesTex · 2026-09-09
- shuding Responds to Highlighting Benchmark Backlash: Prism.js Label Mismatch Explains Gap — shuding · 2026-09-09
- Menlo launches Jan-Nano, a 4B agentic model purpose-built for deep research with MCP — GregoryDiamos · 2026-09-09
- Returning Users Ask: Are ChatGPT Astra's Lowest-Tier Limits Actually Usable? — DragonsRageRS · 2026-09-09
- Anthropic Limits Claude Mythos 5.1 to US Orgs, Cutting Out UK's AISI — Miles_Brundage · 2026-09-09
- Polymarket bets ~93% that OpenAI will preemptively reset Codex weekly limits — Polymarket · 2026-09-09