Skip to content
TheChatArchiveThe Chat Archive
gpt-4-0125-preview· OpenAIAi2 WildChat-4.8M (AI2)

DBRX-Base @databricks also gets to the top 1 on URIAL Bench, which tests Base LLMs (only…

  1. User

    DBRX-Base @databricks also gets to the top 1 on URIAL Bench, which tests Base LLMs (only pre-trained) on the MT-bench with URIAL prompts (three-shot instruction-following IOs). Check out the full results here:

    Check grammar errors

  2. Assistant

    The message is generally clear and correctly conveys the information about DBRX-Base's performance on the URIAL Bench. However, there are a few minor adjustments that can enhance clarity and readability. Here’s a revised version for consideration:

    "DBRX-Base from @databricks has achieved the top position in the URIAL Bench rankings. This benchmark tests base-level Large Language Models (LLMs) in their pre-trained state on the MT-bench, using URIAL prompts which involve three-shot instruction-following input-output tasks. For complete results, check out the full details here:"

    Note: Since the message ends with an invitation to check out the full results but doesn't provide a link or specific direction on where to find these results, it would be beneficial to include a link or specify where or how to access them for completeness.

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