Add Audio-JEPA results and link the Transformers release
Browse files- README.md +3 -2
- export_manifest.json +1 -1
README.md
CHANGED
|
@@ -29,7 +29,7 @@ repository is needed.
|
|
| 29 |
## X-ARES results
|
| 30 |
|
| 31 |
Scores on [X-ARES](https://arxiv.org/abs/2505.16369) (0–100, higher is better).
|
| 32 |
-
BEST-RQ (Conformer), BEST-RQ (ViT),
|
| 33 |
and all BEST-RQ-2 variants reported below are trained on the **same AudioSet
|
| 34 |
split for 200,000 steps**. The pretrained baselines are shown for comparison.
|
| 35 |
|
|
@@ -38,6 +38,7 @@ split for 200,000 steps**. The pretrained baselines are shown for comparison.
|
|
| 38 |
| data2vec | 50.62 | 23.24 | 15.41 | 37.83 | 29.76 | [facebook/data2vec-audio-base](https://huggingface.co/facebook/data2vec-audio-base) |
|
| 39 |
| wav2vec 2.0 | 41.79 | 34.94 | 29.52 | 37.84 | 35.42 | [facebook/wav2vec2-large-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-100k-voxpopuli) |
|
| 40 |
| Whisper | 49.19 | 38.67 | 28.61 | 42.75 | 38.82 | [openai/whisper-base](https://huggingface.co/openai/whisper-base) |
|
|
|
|
| 41 |
| BEST-RQ (Conformer) | 40.43 | 35.58 | 30.81 | 37.43 | 35.60 | Separate codebase |
|
| 42 |
| BEST-RQ (ViT) | 32.87 | 41.62 | 34.50 | 34.88 | 36.33 | [BEST-RQ-ViT](https://huggingface.co/ltuncay/BEST-RQ-ViT) |
|
| 43 |
| BEST-RQ-2 (Interspeech 2026) | 38.49 | 54.40 | 46.39 | 43.21 | 46.43 | [BEST-RQ-2](https://huggingface.co/ltuncay/BEST-RQ-2) |
|
|
@@ -47,7 +48,7 @@ split for 200,000 steps**. The pretrained baselines are shown for comparison.
|
|
| 47 |
Global Mean averages all benchmark task scores. Mean of Means gives equal
|
| 48 |
weight to the Speech, Music, and Environment category means.
|
| 49 |
|
| 50 |
-
|
| 51 |
|
| 52 |
## Model and training
|
| 53 |
|
|
|
|
| 29 |
## X-ARES results
|
| 30 |
|
| 31 |
Scores on [X-ARES](https://arxiv.org/abs/2505.16369) (0–100, higher is better).
|
| 32 |
+
Audio-JEPA, BEST-RQ (Conformer), BEST-RQ (ViT),
|
| 33 |
and all BEST-RQ-2 variants reported below are trained on the **same AudioSet
|
| 34 |
split for 200,000 steps**. The pretrained baselines are shown for comparison.
|
| 35 |
|
|
|
|
| 38 |
| data2vec | 50.62 | 23.24 | 15.41 | 37.83 | 29.76 | [facebook/data2vec-audio-base](https://huggingface.co/facebook/data2vec-audio-base) |
|
| 39 |
| wav2vec 2.0 | 41.79 | 34.94 | 29.52 | 37.84 | 35.42 | [facebook/wav2vec2-large-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-100k-voxpopuli) |
|
| 40 |
| Whisper | 49.19 | 38.67 | 28.61 | 42.75 | 38.82 | [openai/whisper-base](https://huggingface.co/openai/whisper-base) |
|
| 41 |
+
| Audio-JEPA | 29.64 | 44.27 | 25.61 | 31.18 | 33.17 | [Audio-JEPA-base](https://huggingface.co/ltuncay/Audio-JEPA-base) |
|
| 42 |
| BEST-RQ (Conformer) | 40.43 | 35.58 | 30.81 | 37.43 | 35.60 | Separate codebase |
|
| 43 |
| BEST-RQ (ViT) | 32.87 | 41.62 | 34.50 | 34.88 | 36.33 | [BEST-RQ-ViT](https://huggingface.co/ltuncay/BEST-RQ-ViT) |
|
| 44 |
| BEST-RQ-2 (Interspeech 2026) | 38.49 | 54.40 | 46.39 | 43.21 | 46.43 | [BEST-RQ-2](https://huggingface.co/ltuncay/BEST-RQ-2) |
|
|
|
|
| 48 |
Global Mean averages all benchmark task scores. Mean of Means gives equal
|
| 49 |
weight to the Speech, Music, and Environment category means.
|
| 50 |
|
| 51 |
+
Audio-JEPA scores were supplied by the author for [run `jp6l70l6`](https://wandb.ai/tuncay-ludovic/audio%20embeddings/runs/jp6l70l6). The remaining scores are reported in the [project README](https://github.com/LudovicTuncay/audio-embeddings/blob/bb88bf790b1dcf8251c6b38e7a4766534adf33d3/README.md). These are reported research results, not a new benchmark run of the Transformers exports.
|
| 52 |
|
| 53 |
## Model and training
|
| 54 |
|
export_manifest.json
CHANGED
|
@@ -29,7 +29,7 @@
|
|
| 29 |
},
|
| 30 |
"files": {
|
| 31 |
"CODE_LICENSE": "8fe9e8b749cd4abedabcb3100df445db899e72192394d14cc2dbf24a40811af6",
|
| 32 |
-
"README.md": "
|
| 33 |
"adapters.py": "ca0f26826763c0e48b7508da732242bb83adfeb4814e389ed3fde93ad648c728",
|
| 34 |
"config.json": "fe8227548617416bd879efe1be3e9e3fc04270709250f09f01806780fe548789",
|
| 35 |
"configuration_audio.py": "59f3a0b8db0df5af85e677ac33a4431595e9b2eff6d34c1e6a1778dd75a4568d",
|
|
|
|
| 29 |
},
|
| 30 |
"files": {
|
| 31 |
"CODE_LICENSE": "8fe9e8b749cd4abedabcb3100df445db899e72192394d14cc2dbf24a40811af6",
|
| 32 |
+
"README.md": "7039630b0e6bc883e37163ec3d2c383ae2265acd815cdc0b5761e1efc4c92c7a",
|
| 33 |
"adapters.py": "ca0f26826763c0e48b7508da732242bb83adfeb4814e389ed3fde93ad648c728",
|
| 34 |
"config.json": "fe8227548617416bd879efe1be3e9e3fc04270709250f09f01806780fe548789",
|
| 35 |
"configuration_audio.py": "59f3a0b8db0df5af85e677ac33a4431595e9b2eff6d34c1e6a1778dd75a4568d",
|