bosquet.llm.openai-tokens — io.github.zmedelis/bosquet v2023.11.21

JTokkit wrapper to get encode/decode and get token counts. Plus a price estimator for model produced tokens

decode clj

(decode tokens model)

Encode tokens using model

encode clj

(encode text model)

Encode text using model (:gpt-4, :gpt-3.5-turbo)

encoding clj

(encoding model)

Get encoding by model name. Name is provided as keyword matching names specified in

• https://platform.openai.com/docs/models/overview

If model name is not found it will throw NoSuchElementException exception.

fits-in-context-window? clj

(fits-in-context-window? token-count model)

price-estimate clj

(price-estimate prompt model)

(price-estimate prompt completion model)

Estimate price for the prompt and completion using model. If prompt and completion are strings it will count tokens first. If prompt and completion are numbers it will assume they are token counts

pricing clj

OAI model prices (per 1k tokens) and token limits specified in a map:

• input price for the prompt tokens
• output price for the completion tokens
• tokens max context tokens the model supports

registry clj

token-count clj

(token-count text model)

Count tokens in the text using model for token production.