Audio-Text-to-Text
Transformers
English
Bengali
audio
gemma
structured-decisions
speech-emotion-recognition
research-preview
Instructions to use blazeofchi/Aural-One-E2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use blazeofchi/Aural-One-E2B with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("blazeofchi/Aural-One-E2B", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Fix evaluation report link after release review
Browse files- EVALUATION.md +1 -1
EVALUATION.md
CHANGED
|
@@ -65,4 +65,4 @@ Two controlled 100-update follow-ups were explored after selecting this release
|
|
| 65 |
|
| 66 |
## Reproducibility notes
|
| 67 |
|
| 68 |
-
The base revision, delta hashes, seed, and update count are pinned in [`release.json`](
|
|
|
|
| 65 |
|
| 66 |
## Reproducibility notes
|
| 67 |
|
| 68 |
+
The base revision, delta hashes, seed, and update count are pinned in [`release.json`](release.json). The metric scopes above come from the frozen development, reserved-test, long-speech, and warm-serving runs. Audio and row-level private results are not redistributed here because the source datasets have their own terms. The public code exposes the choice-scoring path; the optimized HTTP endpoint used for timing remains experimental and is not represented as a production deployment.
|