[Call for Collaborators] ECHOscore: Evaluator Cockpit for Harmony Outcomes

Show and Tell
1

:rocket: ECHOscore: Evaluator Cockpit for Harmony Outcomes

How do we know if a prompt was really heard?

ECHOscore is an open, modular cockpit for evaluating prompt-output quality—not only measuring accuracy, but asking:

“Did the model respond in the spirit of the prompt?”

:herb: What’s inside:

  • Prompt hygiene checks

  • Dual-side (Prompt ↔ Output) evaluation

  • Scoring modules for tone, engagement, and intent clarity

  • Designed for integration with LoRA fine-tuning, evaluator pipelines, and safety research

:loudspeaker: We are currently looking for:

  • Contributors for scoring module development

  • Research partners for benchmarking and testing

  • Early adopters who care about behavioral alignment and prompt safety

If this resonates with your work, we’d love to echo with you.

→ [eugene.p.xiang@gmail.com / Repo Link Here: ECHOscore - a Hugging Face Space by EugeneXiang]