README.md

metadata

title: ECHOscore
emoji: 📈
colorFrom: blue
colorTo: pink
sdk: gradio
sdk_version: 5.30.0
app_file: app.py
pinned: false
license: cc-by-nc-sa-4.0
short_description: Evaluator Cockpit for Harmony Outcomes

---

⚠️ This is a demo version of ECHOscore.
Data contribution, uploads, and edits are not supported.
To try the full version, please download the desktop release.

🧠 ECHOscore: Evaluator Cockpit for Harmony Outcomes

ECHOscore is a modular evaluation cockpit designed to measure not just the correctness of outputs, but the harmony between prompts and responses—the resonance that defines meaningful alignment.

---

🌟 Why ECHOscore?

ECHOscore is a lightweight, modular evaluation tool for LLM prompts and outputs.

🚀 Features

• ✅ Supports OpenAI, ChatCompletion, LaaJ, DeepEval, OVAL...
• 🧠 Alternative to LangSmith, but simpler and more customizable
• 📦 Plug-and-play scoring rubrics
• 📊 Structured CSV export for QA teams
• 🛠️ Python-native, easy integration with eval.py scripts

💥 ECHOscore = Fast + Flexible + Free

✨ Features

• 🧪 Tri-system scoring: Human, OVAL, and DeepEval
• 📊 Radar chart visualization with Plotly
• 🧠 Stable score hashing for consistent outputs
• 🧼 Network slang annotation (placeholder-ready)
• 💾 Export evaluation results as CSV
• 🧩 Modular architecture for easy customization

---

📦 Modules Overview

Module	Description
app.py	Gradio-based UI interface
eval_core.py	Main orchestrator: scoring, charting, export
scoring_utils.py	Stable scoring + explanation generators
radar_chart.py	Radar chart creation (Plotly)
export_utils.py	CSV generation
config.py	Dimensions, system lists, color map
slang_parser.py	Network slang cleaner (user notes)

---

💡 Philosophy

ECHOscore reflects the principle of human-machine complementarity. It doesn't replace human judgment but amplifies it — allowing humans and models to evaluate collaboratively with clarity, granularity, and oversight.

---

🛡 License

MIT or cc-by-nc-sa-4.0 (as defined in LICENSE file).
For development and commercial usage inquiries, please contact the author.

---

🤝 Acknowledgments

Built with ❤️ by a human-AI co-pilot team.

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference