Post-Training
Also written as Posttraining, Instruction Tuning, Supervised Fine-Tuning, SFT
Everything done to a model after its initial large-scale training to make it useful and safe: teaching it to follow instructions with curated examples (supervised fine-tuning, or SFT), then refining its behaviour with feedback methods such as RLHF, DPO and reinforcement learning. A core research team at every AI lab, and the stage that gives each model its personality and skills.
Think of it like
Pretraining is a broad education; post-training is the job training and coaching that make someone good at a specific role.
Junior or senior?
A genuine lab or model-builder skill, very different from calling a model through an API.
Senior sounds like
Has owned a post-training run end to end — data, training, evaluation — and can describe a trade-off, such as a model getting better at one skill and worse at another.
Ask them
“What did your post-training data look like, and how did you know the new model was actually better?”