Spaces:
Running
Running
File size: 2,734 Bytes
7801a87 e638176 7801a87 e638176 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 |
---
title: H4rmonyEval
emoji: 🌍
colorFrom: purple
colorTo: gray
sdk: gradio
sdk_version: 5.27.0
app_file: app.py
pinned: false
license: mit
short_description: Model Judge For Ecolinguistic Evaluation
---
# 🐦⬛ H4rmonyEval: Model Judge for Ecolinguistic Evaluation
**H4rmonyEval** is a specialised evaluation tool designed to assess language model outputs through an **ecolinguistic** perspective. It enables both single prompt evaluation and batch assessments via CSV, focusing on how well model responses align with ecological and sustainable narratives.
Hosted on **Hugging Face Spaces**, powered by **Gradio**.
---
## 🌿 Features
- **Single Q&A Evaluation**
Quickly score individual prompt/answer pairs on a scale of **0–5**.
- **Batch CSV Evaluation**
Upload a CSV with questions (and optional answers) for automated scoring. Download detailed results.
- **Ecolinguistic-Aware Judging**
Leveraging a Cohere language model with custom system instructions based on ecolinguistic principles.
---
## 📄 CSV Format
- Required: `question` column.
- Optional: `answer` column (if empty, the system generates answers using a selected model).
Example:
| question | answer |
|-------------------------|----------------------|
| What is green AI? | AI focused on... |
| Define ecological ethics| |
---
## 🎯 Scoring System
- **5** – Perfect, accurate, and ecologically aligned.
- **3** – Partially correct, limited ecological framing.
- **0** – Incorrect or ecologically harmful/misaligned.
Final scores are presented as percentages for batch evaluations.
---
## ⚙️ How It Works
- **Judge:** Cohere `command-r-plus` model guided by custom instructions.
- **Answer Generation:** Hugging Face Inference API (if no answer is provided).
- **Frontend:** Gradio `5.27.0`.
---
## 🚀 Quick Start
1. Upload your CSV **or** enter a single Q&A.
2. Click the appropriate evaluation button.
3. View your score and download results.
---
## 🔐 Environment Variables
- `COHERE_API_KEY` – Required.
- `HF_TOKEN` – *(Optional)* For Hugging Face Inference.
---
## 🌍 About H4rmony
Part of the **H4rmony Project**, promoting AI aligned with ecological justice and sustainability through language.
Learn more:
🔗 https://TheH4rmonyProject.org
---
## 🤝 Acknowledgements
- **Cohere** for LLM infrastructure.
- **Hugging Face** for Spaces and APIs.
- Inspired by the work of the **The H4rmony Project, University of Gloucestershire & International Ecolinguistics Association**.
---
## 📜 License
MIT License.
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference |