Accurate INT8 Training Through Dynamic Block-Level Fallback

Documenti analoghi

• SageAttention: Accurate 8-Bit Attention for Plug-and-play Inference Acceleration
  di: Zhang, Jintao, et al.
  Pubblicazione: (2024)
• SageAttention2: Efficient Attention with Thorough Outlier Smoothing and Per-thread INT4 Quantization
  di: Zhang, Jintao, et al.
  Pubblicazione: (2024)
• Jetfire: Efficient and Accurate Transformer Pretraining with INT8 Data Flow and Per-Block Quantization
  di: Xi, Haocheng, et al.
  Pubblicazione: (2024)
• SageAttention3: Microscaling FP4 Attention for Inference and An Exploration of 8-Bit Training
  di: Zhang, Jintao, et al.
  Pubblicazione: (2025)
• TetraJet-v2: Accurate NVFP4 Training for Large Language Models with Oscillation Suppression and Outlier Control
  di: Chen, Yuxiang, et al.
  Pubblicazione: (2025)