Dev builds AES, a non-autoregressive decision head on Qwen2.5-1.5B running in ~28ms batched
CryOrganic8886 · reddit · 2026-09-20
Inspired by TypeSafe AI's Jev, a developer built AES, a proof-of-concept non-autoregressive decision engine on Qwen2.5-1.5B-Instruct: instead of generating tokens, it runs a single prefill pass, passes the final hidden state through a custom FP32 SwiGLU probe, and scores candidate token embeddings by cosine similarity, implementing NOUL/CHOICE/SCORE primitives. Trained on 3,200 samples (BoolQ, ARC, Yelp) for 1 epoch on free Kaggle T4s, it runs in 28ms batched and works on simple multiple-choice tasks with single-token anchors, but has limited zero-shot generalization. Open-sourced on Hugging Face.
More from Research
- minpts-check: one-run predictor tells if your HDBSCAN clusters survive min_samples changes — leland_mcinnes · 2026-11-12
- Architecture study sparks monitoring concerns over a multi-hop planning spike — ChowdhuryNeil · 2026-11-12
- SGLang Megakernel Pushes Muse Spark to 1,109 tok/s, 2.1x Over Baseline — hsu_byron · 2026-11-12
- ArcGrav: a cross-model Agent Control Language to cut token overhead — ArchitecturalGravity · 2026-11-12
- Open-source dual-arm robot solves tangram puzzles with 90 demos, ~60% success — DanGrover · 2026-11-12
- FutureHouse's deslopper trains Kosmos to write indistinguishably from human experts — proteinrosh · 2026-11-12