Instructions to use Reza2kn/visualears-fastconformer-fa-full-ab-fp8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- NeMo
How to use Reza2kn/visualears-fastconformer-fa-full-ab-fp8 with NeMo:
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.ASRModel.from_pretrained("Reza2kn/visualears-fastconformer-fa-full-ab-fp8") transcriptions = asr_model.transcribe(["file.wav"]) - Notebooks
- Google Colab
- Kaggle
| { | |
| "n_eval": 200, | |
| "base_wer": 0.1838006230529595, | |
| "base_cer": 0.06584178498985802, | |
| "base_wall_seconds": 6.194460868835449, | |
| "base_peak_vram_mib": 588.24658203125, | |
| "fp8_wer": 0.1847741433021807, | |
| "fp8_cer": 0.06693711967545639, | |
| "fp8_wall_seconds": 10.259917259216309, | |
| "fp8_peak_vram_mib": 662.06494140625, | |
| "wer_relative_to_base": 1.0052966101694916, | |
| "cer_relative_to_base": 1.0166358595194085, | |
| "wer_retention_vs_base": 0.9947033898305083, | |
| "cer_retention_vs_base": 0.9833641404805915, | |
| "exact_normalized_transcript_match": 0.54, | |
| "exact_matches": 108, | |
| "word_position_agreement_rough": 0.9313197383609081, | |
| "note": "Transcript-level comparison on same 200 FLEURS-fa clips; not CTC-logit argmax parity." | |
| } |