Revisiting MoE and Dense Speed-Accuracy Comparisons for LLM Training.
Xianzhi DuTom GunterXiang KongMark LeeZirui WangAonan ZhangNan DuRuoming PangPublished in: CoRR (2024)
Keyphrases
- training speed
- test set
- high accuracy
- processing speed
- execution speed
- training process
- supervised learning
- prediction accuracy
- computational cost
- high speed
- error rate
- high precision
- precision and recall
- classification accuracy
- three dimensional
- computer vision
- training and testing data
- online learning
- training phase
- training set size
- highly accurate
- real time
- artificial neural networks
- training data
- website
- knowledge base
- neural network