CoolFace
Modelpublic

PrinceAlhassanNasamu/tekyerema-intent-afroxlmr

sourceHugging Facecc-by-nc-4.0updated 4d agoView on Hugging Face
0likes70downloads
Model Card

Intent classifier (AfroXLMR)

Author: Prince Nasamu Alhassan

Overview

270M parameters, and the agent's FAST PATH. It reads Twi, Ewe and Hausa natively — no translation step — and answers in under a millisecond, against roughly 1,140 ms for the 4B student.

Measured intent accuracy .855 Twi, .780 Ewe, .975 Hausa.

It is not the accurate one, it is the quick one. On the same 12 commands per language the 4B student beats it in four languages and ties in two, never losing. Both are kept because they fail differently and the tier system can choose: digit_evidence runs before either, so money and phone numbers are never decided by a model at all.

Use it

No loading snippet for this model yet.

Training data

Trained on the Ghana Speech dataset and related Ghanaian corpora, licensed CC BY-NC 4.0.

Intended use & license

Non-commercial use only (CC BY-NC 4.0). This is inherited from the training data and required by the terms under which the compute was granted: models trained in that window are non-commercial by condition of access, not by inference.

Limitations, stated plainly

  • —Dagbani did get a recogniser, and the claim that it could not was wrong twice over. Every card on this account used to say that "one fine-tuning session on 74 validation rows would not change that". Those 74 rows are the eng-dag machine-translation validation split; the Dagbani speech data in this same account is waxal_dag — 13,228 training rows, 1,750 validation rows, ~71 hours, 1,041 speakers with the largest at 1%. Trained on it, tekyerema-asr-mms-dag scores 36.94 / 11.71, against the 86.59 / 33.95 this project had believed was the ceiling. It still loses to FarmerlineML/w2v-bert-2.0_2026_dagbani_ASR at 29.20 / 9.27, which is what the agent actually serves. A number carried across from a translation table into a speech claim was then repeated on every card here until 2026-09-22.
  • —Evaluation is on read and machine-translated text. No recordings of people speaking agent commands in these languages exist. Numbers measured this way are optimistic about phrasing and pessimistic about code-switching, and should not be read as field performance.
  • —Research work from a hackathon entry, not a supported product.

The rest of the family

Recognisers

Voices

Agent models

Translation

Routing

Acknowledgements

Compute resources provided by AI Skills and Compute Africa (AISCA). Trained on the Ghana NLP H200 GPU. Please keep derivatives non-commercial and share improvements back with the Ghana NLP community (ghananlpcommunity).