Paper Archive

输入关键词搜索 102 篇论文

tracked author

Peiyi Wang

Math-Shepherd first author and DeepSeek-R1 core contributor. OpenReview records Peking University PhD training under Zhifang Sui; GitHub and Hugging Face use the wangpeiyi9979 / peiyi9979 handle family. X search result self-identifies DeepSeek, MathShepherd and DeepSeekMath work.

3 archived notes X: high GitHubX

Related Notes

按论文归档时间排序，展示该作者在本站已经出现的材料。

Jun 23, 2026

DeepSeek V3 Technical Report

DeepSeek V3 用 MLA (Multi head Latent Attention，多头潜变量注意力)、DeepSeekMoE、无辅助损失负载均衡、MTP (Multi Token Prediction，多 token 预测)、FP8 混合精度训练和 DualPipe 通信重叠，把 671B total / 37B active 的开放 MoE (Mixture of Experts，混合专家) 模型训练到强代码、数学和通用...

2412.19437-deepseek-v3-technical-report MoE ArchitectureDistributed TrainingMulti-Token Prediction
Jun 23, 2026

Math Shepherd: Verify and Reinforce LLMs Step by step without Human Annotations

Math Shepherd 的核心价值在于把数学推理步骤的标注问题改写为“当前 step 之后还能否补全到正确答案”的 Monte Carlo potential estimation：对每个中间 step 采样多个 continuation，用最终答案正确性给 step 生成 hard / soft pseudo label，训练 PRM 做 verifier reranking，并进一步把 PRM reward 接入 step b...

2312.08935-math-shepherd-automatic-process-supervision Process SupervisionVerifierReward Modeling
Feb 06, 2026

DeepSeek R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

DeepSeek R1 v2 的核心结论是：大规模 outcome based RL 可以在强 base model 上诱导 long CoT reasoning、自我反思、验证和策略切换等行为；R1 Zero 证明无需 SFT 也能通过 rule based verifiable reward 激发 reasoning capability，R1 则通过 cold start SFT、两阶段 RL、rejection samplin...

2501.12948-deepseek-r1-rl-reasoning Reasoning RLReward ModelingRL Algorithm