AI / Apache Burr Interview questions
What is MapActionsAndStates used for?
MapActionsAndStates runs the full cartesian product of a generator of actions and a generator of states — every action combined with every state — and it is actually the base class that both MapStates and MapActions are built on top of.
class TestModelsOverPrompts(MapActionsAndStates): def actions(self, state, context, inputs): for a in [query_llm.bind(model="gpt-4").with_name("gpt_4_answer"), query_llm.bind(model="o1").with_name("o1_answer")]: yield a def states(self, state, context, inputs): for prompt in ["...", "..."]: yield state.update(prompt=prompt) def reduce(self, state, states): return state.update(all_llm_outputs=[ {"output": s["llm_output"], "model": s["model"], "prompt": s["prompt"]} for s in states ])
Use it when you want to try, say, three prompts against three different models — nine sub-applications in total — rather than varying just one axis at a time. Because every combination becomes its own sub-application, it’s recommended you track which state/action produced each result using values you stash in state itself (like model and prompt above), since they don’t come back automatically labeled.
MapActionsAndStates is the base class for which two other parallelism classes?
Running 3 prompts against 3 models with MapActionsAndStates produces how many sub-applications?
More Related questions...
