ZAYA1-8B Technical Report

Robert Washbourne, Rishi Iyer, Tomas Figliolia, Henry Zheng, Ryan Lorig-Roach, Sungyeon Yang, Pritish Yuvraj, Quentin Anthony, Yury Tokpanov, Xiao Yang, Ganesh Nanduru, Stephen Ebert, Praneeth Medepalli, Skyler Szot, Srivatsan Rajagopal, Alex Ong, Bhavana Mehta, Beren Millidge

arXiv:2605.05365 · 2026-05-08 공개 · arXiv · PDF

Abstract

We present ZAYA1-8B, a reasoning-focused mixture-of-experts (MoE) model with 700M active and 8B total parameters, built on Zyphra's MoE++ architecture. ZAYA1-8B's core pretraining, midtraining, and supervised fine-tuning (SFT) were performed on a full-stack AMD compute, networking, and software platform. With under 1B active parameters, ZAYA1-8B matches or exceeds DeepSeek-R1-0528 on several challenging mathematics and coding benchmarks, and remains competitive with substantially larger open-weight reasoning models. ZAYA1-8B was trained from scratch for reasoning, with reasoning data included from pretraining onward using an answer-preserving trimming scheme. Post-training uses a four-stage RL cascade: reasoning warmup on math and puzzles; a 400-task RLVE-Gym curriculum; math and code RL with test-time compute traces and synthetic code environments built from competitive-programming references; and behavioral RL for chat and instruction following. We also introduce Markovian RSA, a test-time compute method that recursively aggregates parallel reasoning traces while carrying forward only bounded-length reasoning tails between rounds. In TTC evaluation, Markovian RSA raises ZAYA1-8B to 91.9\% on AIME'25 and 89.6\% on HMMT'25 while carrying forward only a 4K-token tail, narrowing the gap to much larger reasoning models including Gemini-2.5 Pro, DeepSeek-V3.2, and GPT-5-High.

한국어 요약

📋 한 줄 요약

**[LLM 사전학습/추론]** AMD 하드웨어에서 처음부터 추론용으로 학습된 8B MoE 모델 ZAYA1-8B와 4단계 RL 후처리, Markovian RSA 테스트타임 압축 기법을 공개한다.

🎯 핵심 기여도

💡 핵심 아이디어

추론 능력은 사전학습 단계부터 데이터 분포에 포함시키고, RL은 단계별 커리큘럼으로 강화하며, 추론 시점에는 평행 추적을 짧은 꼬리로 압축해 다음 라운드에 전달함으로써 메모리 비용 없이 테스트타임 컴퓨트를 확장한다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: 비-NVIDIA 스택에서 SOTA급 추론 모델 학습이 가능함을 실증, 효율적 MoE+TTC 조합의 강력함 입증. **한계**: 추론 외 일반 능력에 대한 평가가 제한적이며 Markovian 가정이 깨지는 장기 추론에서의 성능은 미검증.

🚀 실용적 활용