Either by adding multiple LM heads ([medusa](https://arxiv.org/abs/2401.10774)) or using a drafter model. Alternative: https://github.com/SafeAILab/EAGLE Alternative: https://arxiv.org/html/2502.09419v1
Either by adding multiple LM heads (medusa) or using a drafter model.
Alternative: https://github.com/SafeAILab/EAGLE
Alternative: https://arxiv.org/html/2502.09419v1