AI Architecture & Systems

The Full Lifecycle of a Foundation Model

6.2.1Pretraining, Continued Pretraining, and Mid-Training#

6.2.2Supervised Fine-Tuning and Instruction Tuning#

6.2.3Preference Optimization and Reinforcement Learning#

6.2.4Distillation, Compression, and Deployment#

6.2.5Inference, Test-Time Compute, and Continuous Evaluation#

6.2.6Scaling Laws and Compute-Optimal Training#