A dock office only needs a three-way tag on each radio transcript and must stay fast and on-device. Why prefer a small encoder classifier over a large generative LLM?
Select an answer to reveal the explanation.
Short Explanation
The dock office only needs a three-way tag, fast and on-device. When the answer is a label, a span, or a vector, a small encoder is cheaper and easier to control than a big generator writing open-ended prose. A decoder is not required, and NCCL is not the on-device choice.
Full Explanation
A closed tag is an encoder job: smaller, faster, and more controlled than calling a large generator for open-ended prose. A decoder is not required, and NCCL is not the on-device choice.