Hermes Escalation v1
A Laya typed-decision model, fine-tuned with System One Studio on an Apple silicon Mac. It answers the questions below in a single forward pass, with calibrated probabilities and zero generated tokens.
Base model: convaiinnovations/laya · method: lora, proper objective · trained in
29 minutes on an Apple silicon Mac.
Measured on the held-out test split
| Metric | Base model | This model |
|---|---|---|
| Accuracy | 23.3% | 76.7% [59.1%–88.2%] |
| Calibration error (ECE) | 0.518 | 0.188 |
| Log loss | 1.594 | 0.569 |
| Brier score | 0.773 | 0.213 |
| Decisions scored | 30 | 30 |
Fine-tuning fixed 20 test decisions the base model got wrong and broke 4 it got right (exact McNemar p = 0.002).
Test rows were never trained on. Accuracy intervals are Wilson intervals; the paired test is an exact McNemar test between the base and the fine-tuned model on the same rows.
Use it
pip install systemonemodels
systemone pull vorthalis/hermes-escalation-v1
import json
import laya_mlx as laya # pip install laya-mlx, on Apple silicon
from systemone import snapshot_download
path = snapshot_download("vorthalis/hermes-escalation-v1")
agent = laya.load(str(path))
questions = json.loads((path / "questions.json").read_text())
print(agent.predict("your text here", questions)["answers"])
Ask it these questions: the instructions and option texts are part of the model's input, so changing them changes the task it was tuned for.
{
"needs_human": {
"type": "noul",
"instructions": "Requires human intervention"
},
"retry_strategy": {
"type": "choice",
"instructions": "Select retry strategy",
"criteria": [
"retry_same",
"switch_model",
"decompose",
"abort"
]
}
}
The same folder also loads in the upstream PyTorch package on Linux and NVIDIA, and System One Studio can export it to ONNX.