bosquet.complete — io.github.zmedelis/bosquet v2023.11.21

->cache clj

(->cache)

cache clj

chat-completion clj

(chat-completion messages
                 {{service wkk/service model-params wkk/model-parameters}
                    llm.chat/conversation})

complete clj

(complete prompt {gen-var wkk/gen-var-name :as opts})

evict clj

(evict prompt model-params)

generate-with-cache clj

(generate-with-cache cache? generator prompt model-parameters)

Call generator function with prompt and model-parameters. If cache? is true then use cache store previously generated result.

Cache gets a hit if prompt and model-parameters are the same.