Vektor MLX Models
Collection
6 items • Updated
How to use Nocti-Dev/nb-whisper-large-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download Nocti-Dev/nb-whisper-large-mlx --local-dir nb-whisper-large-mlx
MLX conversion of NbAiLab/nb-whisper-large at commit 8c6249fdeeb4dcd05e5735a4c39640607eb6e4ac, made with Prism. Self-contained: weights, config and tokenizer are all in this repo, so it loads offline.
| Source | NbAiLab/nb-whisper-large @ 8c6249fdeeb4 |
| Recipe | whisper-hf v1 |
| dtype | float16 |
| Quantization | none |
| Licence | apache-2.0 (from NbAiLab/nb-whisper-large) |
| Worker | prism-worker 0.1.0, Python 3.14.10, mlx 0.32.4, numpy 2.5.4, huggingface_hub 2.3.0, hf-xet 1.7.0, torch 2.14.1, transformers 5.21.0 |
| Worker lockfile | 6f31327071399b5c (sha256) |
The unconverted model (NbAiLab/nb-whisper-large@8c6249f, transformers 5.21.0 / torch 2.14.1) and this conversion (mlx-audio-swift@dbe5eaa) transcribed the same short Norwegian test clip: passed, the transcripts differ by a word error rate of 0.00 (max 0.10).
| Word error rate | |
|---|---|
| Converted vs unconverted (verdict) | 0.00 |
| Converted vs script | 0.46 |
| Unconverted vs script | 0.46 |
| File | Size | sha256 |
|---|---|---|
added_tokens.json |
34,648 | 3c51f66c4c21f9e126970078f11ae77a78c74aee8df606ee9daba86e467108e0 |
config.json |
3,627 | 7aa39e298a0cf538f31375fbfe6a5c752f672f059b7b231dcb72689174a24908 |
generation_config.json |
3,950 | 5cbdba82fb3588b71ef0a58e6d7489e2350bdfb75725636759e3459a3a523c42 |
merges.txt |
493,869 | 2df2990a395e35e8dfbc7511e08c12d56018d8d04691e0133e5d63b21e154dc6 |
model.safetensors |
3,087,130,926 | dcfe4ebda15877d5fb5868ee883fb3a533b65082b47de307a32b243105f20401 |
preprocessor_config.json |
340 | 7ccc62c6f2765af1f3b46c00c9b5894426835a05021c8b9c01eecb6dfb542711 |
special_tokens_map.json |
2,186 | baea4ea09372eb4fca86b4e4346139fd73cb807d5087e9de0948e971739c3e74 |
tokenizer.json |
2,481,355 | 5ceb4ad41152dcbf89d26884faaa71c04e2f0a12b2d2c88a12894eee1e55c189 |
tokenizer_config.json |
282,860 | 452bb0ea53109970b77865f5b6dcc329c6fa255e3bc9eec6ae143b120be5fa57 |
vocab.json |
835,528 | 6788c80b082e9b0d1393147d3a3e62ba19285ac0c82ace8e5ef00f37ead58971 |
Full provenance is in prism.json.
Quantized