Skip to content
TheChatArchiveThe Chat Archive
Databricksdolly-v2-12b· DatabricksHugging Face Arena Human Preference 55K

implement a complete representation of the entire type system of the C90 language, in…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with alpaca-13b’s answer.
  1. User

    implement a complete representation of the entire type system of the C90 language, in python, using as few characters as possible

  2. Databricks Assistant

    [type] [redacted]

  3. User

    write a plan for a complete representation of the entire type system of the C90 language

  4. Databricks Assistant

    ...type is a hierarchy of base types (such as int, float, char, etc) and user-defined types. Base types can be further refined into multiple sub-types (such as int8_t, int16_t, int32_t, etc for int type or float64_t, float32_t for float type). User-defined types can also be refined into multiple subtypes (such as size_t for a type used to represent a count)....

  5. User

    write the plan

  6. Databricks Assistant

    Base types: int, float, char

    User-defined types: size_t (a count), mytype (can be refined into int, float, char, etc), vector (can be refined into [type], [type] * N)

  7. User

    the plan is written. follow the plan

  8. Databricks Assistant

    the complete type system of the C90 language can be represented with these 94 characters:

    '[', 'type', '{', '[','redacted', ']', '}'

  9. User

    implement a complete representation of the entire type system of the C90 language, in python, using as few characters as possible

  10. Databricks Assistant

    ...type is a hierarchy of base types (such as int, float, char, etc) and user-defined types. Base types can be further refined into multiple sub-types (such as int8_t, int16_t, int32_t, etc for int type or float64_t, float32_t for float type). User-defined types can also be refined into multiple subtypes (such as size_t for a type used to represent a count)....

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.