Skip to content
System One

Jev vs Laya on Moroccan Darija sentiment

A zero-shot benchmark of Jev, self-hosted Kev-9B and self-hosted Laya on 171 Moroccan Darija reviews in Arabic script and Arabizi. Jev scored 79.5 percent accuracy, Kev-9B 71.9 percent, and Laya 53.8 percent, tying an always-positive baseline.

Upstream djev-run Snake demo with board, move probabilities, and live decision payload
Image from github.com

mouadse/jev-vs-laya on GitHub

Stars
0
Forks
0
Language
Python
Last push
23 Sep 2026

Read from GitHub on . Counts change daily.

How it uses Jev

def predict(self, text: str) -> Prediction:
    started = time.perf_counter()
    try:
        response = self._client.system_one(
            state={"review": text},
            questions={"sentiment": self._question},
        )
    except _TRANSIENT as error:
        raise TransientBackendError(_safe_error(error)) from error
    except _PERMANENT as error:
        raise PermanentBackendError(_safe_error(error)) from error
    elapsed_ms = (time.perf_counter() - started) * 1000
    return parse_response(response, elapsed_ms)

def close(self) -> None:

View in src/darija_eval/backends/jev.py

Open on GitHub

More like this

Keep browsing