Jina-OCR-v1：以推測式解碼與稠密可驗證獎勵做高效文件解析
Jina-OCR-v1: Efficient Document Parsing with Speculative Decoding and Dense Verifiable Rewards
鄭子凡 許願
MiMo-V2.6：將強化學習擴展至自我改進
MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement
Liang-Hsun Huang 許願
稀疏再造 (Sparse Upcycling)：從稠密檢查點訓練 Mixture-of-Experts
Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints
Liang-Hsun Huang 許願
MiniMax-M1：以閃電注意力高效地擴展測試時運算
MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention
Liang-Hsun Huang 許願
MiniCPM：以可擴展的訓練策略揭示小型語言模型的潛力
MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies
Liang-Hsun Huang 許願
Muon 可以擴展到 LLM 訓練
Muon is Scalable for LLM Training
Liang-Hsun Huang 許願
Kimi K3：開放的前沿智慧
Kimi K3: Open Frontier Intelligence
Liang-Hsun Huang 許願
論 Qwen3.8-Next 架構設計：評測、效率與訓練穩定性
On the Design of Qwen3.8-Next Architecture: Evaluation, Efficiency, and Training Stability
Liang-Hsun Huang 許願
MoE 的持續預訓練：你的路由器有多穩健？
Continual Pre-training of MoEs: How robust is your router?
Liang-Hsun Huang 許願
預訓練 LLM 的原地詞元化器擴充
In-Place Tokenizer Expansion for Pre-trained LLMs
Liang-Hsun Huang 許願
指令遵循式剪枝：面向大型語言模型
Instruction-Following Pruning for Large Language Models
林伯燊 許願
Nemotron 3 Nano：開放高效的 MoE 混合 Mamba-Transformer 智能體推理模型
Nemotron 3 Nano Open Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
林伯燊 許願
Nemotron 3 Super：開源高效混合專家 Mamba-Transformer 模型，面向智能體推理
Nemotron 3 Super Open Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
Teds Lin 許願
Qwen3 技術報告
Qwen3 Technical Report
Liang-Hsun Huang 許願