|
|
--- |
|
|
license: cc-by-nc-sa-4.0 |
|
|
tags: |
|
|
- multilingual |
|
|
- semantic-graph |
|
|
- cognitive-diversity |
|
|
- agent-alignment |
|
|
datasets: |
|
|
- AI-Research-Agent-Team/LIMIT-GRAPH-v1.1 |
|
|
language: |
|
|
- en |
|
|
- id |
|
|
- ar |
|
|
- es |
|
|
--- |
|
|
|
|
|
# 🧠 DCoTAgentAligner Base Model |
|
|
|
|
|
This model aligns diverse cognitive reasoning styles across multilingual corpora using semantic graph vocabularies. It supports English, Indonesian, Arabic, and Spanish. |
|
|
|
|
|
## 🔍 Capabilities |
|
|
|
|
|
- Chain-of-thought generation with style control |
|
|
- Semantic graph alignment via LIMIT-GRAPH |
|
|
- Multilingual adaptation with cultural hooks |
|
|
- Evaluation-ready for leaderboard submission |
|
|
|
|
|
## 📊 Benchmark Results |
|
|
|
|
|
| Language | Accuracy | Diversity Index | Graph Consistency | |
|
|
|----------|----------|-----------------|-------------------| |
|
|
| English | 0.87 | 0.78 | 0.82 | |
|
|
| Indonesian | 0.85 | 0.75 | 0.80 | |
|
|
| Arabic | 0.83 | 0.72 | 0.79 | |
|
|
|
|
|
## 📁 GitHub + Leaderboard |
|
|
|
|
|
- [GitHub Repository](https://github.com/AI-Research-Agent-Team/ai_research_agent_DCoTAgentAligner) |
|
|
- [LIMIT-GRAPH Leaderboard](https://github.com/AI-Research-Agent-Team/LIMIT-GRAPH#leaderboard) |
|
|
|
|
|
## 🧪 Evaluation Harness |
|
|
|
|
|
Use `benchmark.py` from the repo to validate your agent before submission. |
|
|
|
|
|
## 📜 License |
|
|
|
|
|
This model is licensed under [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/). |
|
|
|
