Instructions to use Reson8/Rumble-2-Flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Reson8/Rumble-2-Flash with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="Reson8/Rumble-2-Flash")# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("Reson8/Rumble-2-Flash") model = AutoModelForMultimodalLM.from_pretrained("Reson8/Rumble-2-Flash", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Add Open ASR Leaderboard results (English and multilingual)
Browse filesResults from the Open ASR Leaderboard, English and multilingual tracks, with RTFx.
README.md
CHANGED
|
@@ -130,3 +130,47 @@ Pass these options through vLLM's `chat_template_kwargs`, or directly to
|
|
| 130 |
## License
|
| 131 |
|
| 132 |
Licensed under the [Apache License 2.0](LICENSE).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
## License
|
| 131 |
|
| 132 |
Licensed under the [Apache License 2.0](LICENSE).
|
| 133 |
+
|
| 134 |
+
## Open ASR Leaderboard results
|
| 135 |
+
|
| 136 |
+
Evaluated on the [Open ASR Leaderboard](https://huggingface.co/spaces/hf-audio/open_asr_leaderboard?version=1.16.1) (release 18-11-2026, v1.16.1), run on an H200 with vLLM. Compare against other open models on the multilingual tracks with [this view](https://huggingface.co/spaces/hf-audio/open_asr_leaderboard?languages=Reson8%2FRumble-2-Flash&proprietary=false) (API models hidden).
|
| 137 |
+
|
| 138 |
+
### English track
|
| 139 |
+
|
| 140 |
+
Average WER over the 7 default splits: **6.08** (RTFx 6338).
|
| 141 |
+
|
| 142 |
+
| Split | WER (%) | RTFx |
|
| 143 |
+
|---|---|---|
|
| 144 |
+
| AMI | 9.24 | 4245 |
|
| 145 |
+
| Earnings22 | 6.45 | 4811 |
|
| 146 |
+
| GigaSpeech | 7.56 | 5978 |
|
| 147 |
+
| SPGISpeech | 3.08 | 7308 |
|
| 148 |
+
| URGENT24 | 10.25 | 3314 |
|
| 149 |
+
| Voice Arena Monsoon | 3.56 | 3760 |
|
| 150 |
+
| VoxPopuli | 2.43 | 4222 |
|
| 151 |
+
|
| 152 |
+
### Multilingual track
|
| 153 |
+
|
| 154 |
+
| Language | Dataset | WER (%) | RTFx |
|
| 155 |
+
|---|---|---|---|
|
| 156 |
+
| German | FLEURS | 4.77 | 5967 |
|
| 157 |
+
| German | Common Voice | 6.28 | 5967 |
|
| 158 |
+
| Spanish | FLEURS | 2.82 | 6079 |
|
| 159 |
+
| Spanish | Common Voice | 4.33 | 6079 |
|
| 160 |
+
| Spanish | MLS | 4.04 | 6079 |
|
| 161 |
+
| French | FLEURS | 5.95 | 6001 |
|
| 162 |
+
| French | Common Voice | 10.01 | 6001 |
|
| 163 |
+
| French | MLS | 5.66 | 6001 |
|
| 164 |
+
| Italian | FLEURS | 2.53 | 5772 |
|
| 165 |
+
| Italian | Common Voice | 4.84 | 5772 |
|
| 166 |
+
| Italian | MLS | 7.90 | 5772 |
|
| 167 |
+
| Dutch | FLEURS | 6.05 | 3494 |
|
| 168 |
+
| Dutch | Common Voice | 4.92 | 3494 |
|
| 169 |
+
| Dutch | MLS | 9.38 | 3494 |
|
| 170 |
+
| Portuguese | FLEURS | 5.31 | 4723 |
|
| 171 |
+
| Portuguese | MLS | 7.61 | 4723 |
|
| 172 |
+
| Hindi | Voice Arena Monsoon | 14.11 | 2101 |
|
| 173 |
+
| Chinese (CER) | FLEURS | 5.82 | 5379 |
|
| 174 |
+
| Chinese (CER) | AISHELL-1 | 7.31 | 5379 |
|
| 175 |
+
|
| 176 |
+
RTFx is the inverse real-time factor (seconds of audio transcribed per second of compute).
|