Views
No views yet
| File | Bytes |
|---|---|
model.pt | 12,325,897,883 |
omniASR_tokenizer_written_v2.model | 91,481 |
card.yaml | — |
requirements.txt | — |
omnilingual-asr checkpoint, loaded through the asset
card shipped alongside it (card.yaml). The solution repository wraps the whole
contract -- pinned environment, batch-size-1 decode, tokenizer wiring -- in one
CLI:1git clone https://github.com/DariusTheGeek/waxal-asr-solution
2cd waxal-asr-solution && bash install.sh # pinned environments, ~15 min
3python models/download_models.py --repo waxal-lin-omniasr-ctc-3b
4
5.venvs/omni/bin/python inference/decode/omniasr.py \
6 --config configs/lin/ctc3b.yaml \
7 --audio path/to/wav_dir --output transcripts.csv--weights accepts any directory holding this repo's files, e.g. the path
returned by huggingface_hub.snapshot_download("DariusTheGeek/waxal-lin-omniasr-ctc-3b").
requirements.txt in this repo pins the runtime alone; the environment lock
the release was verified under is env/requirements-omni.txt in the solution
repository.| Parent | facebook/omniASR-CTC-3B-v2 |
| Language | Lingala |
| Fine-tuning data | Waxal Lingala/Shona supervised split (google/WaxalNLP) |
| Seed | 42 |
https://github.com/DariusTheGeek/waxal-asr-solution for the environment locks, decode configuration and the exact
command that reproduces the submission end to end from audio.apache-2.0, inherited from the parent model.