Round 9: Tossup 18

The number of these things is the numerator of a ratio that is empirically optimal at about 20, the Chinchilla point. Computations involving the Q, K, and V vectors generated from these things are stored in a KV cache, which improves a metric called the “time to the first” one of these things. The process of converting an input into these things is commonly done using byte-pair encoding. During the prefill stage, IDs corresponding to these things are embedded into a vector that is then passed through multiple transformer layers. The number of these things that can be processed at once is the context window. The cost of AI compute (10[1])is often (-5[1])reported as the price (10[1])per (10[1])one million of these (10[1])things. (-5[1])For (10[1])10 points, name these units of text that large language models (10[1])split queries into and try to predict the next one of. ■END■ (10[1]0[2])

Back to tossups

Buzzes

Summary