Beyond Tool Access — mcAgent v20
Verdict
Fine-tuning earns its claim beyond tool access alone, on both evaluation sets.
The base model essentially never emits a working tool call at all, regardless of phrasing. Tool access alone does not confer tool-use behavior at this model size.
Held-out set — n = 106
- Tool-call rate
- 0% rises to 98%
- Answer rate
- 48% rises to 96%
Adversarial probe — n = 43
- Tool-call rate
- 2% rises to 83%
- Answer rate
- 13% rises to 86%
base model → fine-tuned v20