cy0307/ag-agent-harness

Agent + tool-use harness

A reusable harness: a tool registry, a tool-using agent loop, a task suite, and a scorer.

Trained from scratch in Ropedia Academy — an interactive, bilingual course on embodied & spatial AI. Educational model: small and quick to train; the value is the method and a reproducible pipeline, not a leaderboard score. Try it live in the Ropedia demos Space.

At a glance

Base model	Trained from scratch (random initialization) — no pretrained base model.
Task	agent evaluation
Training objective	No training — a tool-use agent loop + scorer (an evaluation harness).
Track	AG · Agents & RL
Notebook

Dataset

• Name: Tool-use task suite
• Type: synthetic — procedural
• Size / stats: 5 graded tasks (arithmetic + string ops) with ground-truth answers
• Split: eval suite
• Source: procedural

Training config

No training — a tool registry + ReAct-style loop + scorer over a 5-task suite.

Evaluation results

metric	value	meaning
success_rate	1.0	fraction of agent tasks solved with the correct final answer

Inference example

import torch
state = torch.load("model.pt", map_location="cpu")   # this repo's checkpoint
# Rebuild the exact module from the lab notebook (see "Reproduce"), then:
# model.load_state_dict(state); model.eval()

Limitations

Educational scale. Trained quickly on CPU on small or synthetic data, so absolute numbers are not competitive with production systems — the value is the method and a reproducible pipeline. No large-scale data, no hyperparameter sweep, and no multi-seed variance is reported. Not for production use.

Failure cases

Only as strong as the task suite & scorer; brittle tool-call parsing; doesn't probe long-horizon planning.

Reproduce / train your own

One click: open the notebook in Colab → Runtime → GPU → Run all, then run its Publish to the Hugging Face Hub cell.

From a shell:

git clone https://github.com/ChaoYue0307/ropedia-academy.git && cd ropedia-academy
pip install torch numpy matplotlib scikit-learn scikit-image gymnasium
jupyter nbconvert --to notebook --execute notebooks/training/AG_agent_harness.ipynb --output run.ipynb
# optional: override training length, e.g.  STEPS=2000  (or EPISODES=600)  before running

Files

• figure.png
• results.json

License

Code & weights: MIT (this repository) — educational use encouraged.
Data: generated procedurally in the notebook — no external dataset.

Citation

If you use this model or the course materials, please cite:

@misc{ropedia_academy,
  title  = {Ropedia Academy: an interactive course on embodied & spatial AI},
  author = {Ropedia Academy},
  year   = {2026},
  howpublished = {\url{https://chaoyue0307.github.io/ropedia-academy/}}
}

Method / original work: Yao et al., ReAct, ICLR 2023; Schick et al., Toolformer, 2023.

Related assets

• 🚀 Live demos: https://huggingface.co/spaces/cy0307/ropedia-demos
• 🤗 All trained models + collection: https://huggingface.co/cy0307
• 📚 Course & all labs: https://chaoyue0307.github.io/ropedia-academy/ · Labs tab
• 💻 Source / notebooks: github.com/ChaoYue0307/ropedia-academy

---

Part of the Ropedia Academy trained-model collection. Contributions & issues welcome on GitHub.