PrinceAlhassanNasamu/tekyerema-1-tool

Tɛkyerɛma-1 tool adapter (arm 1)

Author: Prince Nasamu Alhassan

Overview

Qwen3-4B + LoRA. Picks one of 21 phone tools from an utterance. Trained on router_input — the text AFTER an eng→X→eng round trip — so it learns to recover intent from translation damage.

Arm ① against arm ②, measured 2026-08-31. 60 commands whose text appears nowhere in training (139 candidates skipped for repeating it), each arm reading the input it was trained for:

language	arm ① (translate, then route)	arm ② (native)
Kusaal	6/12	9/12
Ewe	13/13	12/13
Dagbani	8/8	8/8
Ghanaian English	10/10	10/10
Hausa	7/7	7/7
Twi	10/10	10/10
total	54/60 · 90.0%	56/60 · 93.3%

Removing the translator costs nothing in five languages and gains three points in Kusaal — the language whose translation is weakest, which is exactly where the argument for arm ② predicted it would matter.

The comparison is text-only, and both sides read machine-translated corpus text, so it measures ROUTING rather than the spoken pipeline. On this evidence arm ② is the one to serve.

Use it

A LoRA adapter: load the base, then apply it. Take the tokenizer from the base, not from this repo — the adapter's saved tokenizer carries a chat template that silently ignores enable_thinking, and without that flag Qwen3 opens a reasoning block and never reaches the JSON.

import torch
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

BASE = "Qwen/Qwen3-4B"
tok  = AutoTokenizer.from_pretrained(BASE)         # the BASE, deliberately
model = PeftModel.from_pretrained(
    AutoModelForCausalLM.from_pretrained(
        BASE, dtype=torch.bfloat16, device_map="auto"),
    "PrinceAlhassanNasamu/tekyerema-1-tool").eval()

prompt = tok.apply_chat_template(
    [{"role": "user", "content": TOOL_PROMPT}],    # schema + user command
    add_generation_prompt=True, tokenize=False,
    enable_thinking=False)                         # not optional
enc = tok(prompt, return_tensors="pt").to(model.device)
out = model.generate(**enc, max_new_tokens=96, do_sample=False)
print(tok.decode(out[0][enc["input_ids"].shape[1]:],
                 skip_special_tokens=True))

TOOL_PROMPT must carry the full 21-tool schema exactly as training did — see H200/sessions/s3_train_tekyerema1.py. Asked without it, the model has no tool names to choose from and scores zero.

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 had no recogniser of its own for this whole project, and the reason given for that was wrong. Every card here said "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% — more data and better speaker diversity than Ewe, which produced a working 42.19 WER recogniser. A number was carried across from a translation table into a speech claim, and then repeated on every model card on the account. It is training now, on 2026-08-31. Until it is scored, the honest statement is that Dagbani's best available recogniser scores 86.6 WER and nobody had tried fine-tuning on the data already in hand.
• 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

• whisper-large-v3-turbo-tekyerema-eng-foundation — Ghanaian English ASR — course 1 (foundation)
• kusaal-whisper-small-lora — Kusaal ASR (Whisper-small LoRA, superseded)
• kasa42-asr — KASA-42 (Kusaal, third-party export)
• tekyerema-asr-ctc — Twi ASR (w2v-BERT CTC)
• tekyerema-asr-mms-ewe — Ewe ASR (MMS adapter)
• tekyerema-asr-mms-dag — Dagbani ASR (MMS adapter)
• tekyerema-asr-mms-hau — Hausa ASR (MMS adapter)
• tekyerema-asr-mms-kus — Kusaal ASR (MMS adapter)
• whisper-large-v3-turbo-tekyerema-eng — Ghanaian English ASR (Whisper large-v3-turbo)

Voices

• tekyerema-tts-twi — Twi TTS (VITS)
• tekyerema-tts-kus — Kusaal TTS (VITS)
• tekyerema-tts-ewe — Ewe TTS (VITS)
• tekyerema-tts-hau — Hausa TTS (VITS)
• tekyerema-tts-eng — Ghanaian English TTS (VITS)

Agent models

• tekyerema-1-reply — Tɛkyerɛma-1 reply adapter (arm ①)
• tekyerema-1-native-reply — Tɛkyerɛma-1 reply adapter (arm ②)
• tekyerema-1-tool — Tɛkyerɛma-1 tool adapter (arm 1)
• tekyerema-audio-native — Tɛkyerɛma-1 audio-native (arm 3)
• tekyerema-audio-native-4k — Tɛkyerɛma-1 audio-native, 4,000 clips (arm 3 v2)
• tekyerema-1-native-tool — Tɛkyerɛma-1 tool adapter (arm 2)

Translation

• tekyerema-nllb600m-v1 — Tɛkyerɛma MT v1 (NLLB-600M)
• kusaal-nllb-600M — Kusaal MT specialist (NLLB-600M)

Routing

• tekyerema-intent-afroxlmr — Intent classifier (AfroXLMR)

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).