3f:I[185,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/1e2btqyhcbvr0.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/1pfyc3q3c61jb.js","/_next/static/chunks/35-gbipsuw5b6.js"],"SendEmailButton"] 41:I[63625,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/1e2btqyhcbvr0.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/1pfyc3q3c61jb.js","/_next/static/chunks/35-gbipsuw5b6.js"],"AddToLibraryButton"] 42:I[29721,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/1e2btqyhcbvr0.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/1pfyc3q3c61jb.js","/_next/static/chunks/35-gbipsuw5b6.js"],"ShareButton"] 43:I[58016,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/1e2btqyhcbvr0.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/1pfyc3q3c61jb.js","/_next/static/chunks/35-gbipsuw5b6.js"],"ProcessEpisodeButton"] 44:I[45269,["/_next/static/chunks/3xnt26d00_zk-.js","/_next/static/chunks/1e2btqyhcbvr0.js","/_next/static/chunks/42spz6paskvb6.js","/_next/static/chunks/06x30xgfe6iur.js","/_next/static/chunks/1pfyc3q3c61jb.js","/_next/static/chunks/35-gbipsuw5b6.js"],"SignupCTA"] 39:["$","div",null,{"className":"flex flex-wrap items-center gap-x-4 gap-y-1 font-mono text-[12px] uppercase tracking-[0.1em] text-[color:var(--fg-3)]","children":[["$","time",null,{"dateTime":"2026-07-21T03:52:15.000Z","children":"July 21, 2026"}],["$","span",null,{"aria-hidden":true,"children":"·"}],["$","span",null,{"children":"00:19:35"}],false]}] 40:T4c1,

🤗 Upvotes: 56 | cs.CL, cs.AI

Authors:
Zitian Gao, Yilong Chen, Yihao Xiao, Xinyu Yang, Ran Tao, Joey Zhou, Bryan Dai

Title:
Loop the Loopies!

Arxiv:
http://arxiv.org/abs/2607.16051v2

Abstract:
We present the Loopie series, consisting of two Mixture-of-Experts (MoE) models: a 20B-parameter model with 2B active parameters and a 6B-parameter model with 0.6B active parameters. Looped Transformers have long faced a challenge: given an N times increase in pre-training compute, increasing the parameter count by a factor of N usually outperforms looping a model N times. Loopie addresses this challenge. Extensive ablation studies, including comparisons with a vanilla 30B-A3B model, show that Loopie substantially outperforms vanilla Transformer baselines trained with the same compute budget. With a novel post-training method, Loopie develops strong reasoning abilities and achieves frontier-level reasoning performance.