TST — Token Superposition Training

Nous Research unveiled Token Superposition Training (TST), a technique claiming 2.5x faster LLM training times by optimizing how models process redundant data sequences. TST works by superimposing tokens during training to reduce compute on repetitive patterns, with zero architectural changes required to the final deployed model.

Nous Research unveiled Token Superposition Training (TST), a technique claiming 2.5x faster LLM training times by optimizing how models process redundant data sequences. TST works by superimposing tokens during training to reduce compute on repetitive patterns, with zero architectural changes required to the final deployed model.

• 012.5x faster LLM pretraining with no changes to the final model architecture
• 02Token superposition reduces compute on redundant sequence patterns
• 03Compatible with standard transformer architectures and training pipelines