Update README.md
Browse files
README.md
CHANGED
@@ -6,6 +6,7 @@ inference: false
|
|
6 |
# Introduction
|
7 |
|
8 |
**MERT-v0** is a completely unsupervised model trained on 1000 hour music audios.
|
|
|
9 |
It is SOTA-comparable on multiple MIR tasks even under probing settings, while keeping fine-tunable on a single 2080Ti.
|
10 |
It outperforms Jukebox representation on GTZAN (genre classification) and GiantSteps (key classification) datasets.
|
11 |
Larger models trained with more data are on the way~
|
|
|
6 |
# Introduction
|
7 |
|
8 |
**MERT-v0** is a completely unsupervised model trained on 1000 hour music audios.
|
9 |
+
Its architecture is similar to the [HuBERT model](https://huggingface.co/docs/transformers/model_doc/hubert), but it has been specifically designed for music through the use of specialized pre-training strategies.
|
10 |
It is SOTA-comparable on multiple MIR tasks even under probing settings, while keeping fine-tunable on a single 2080Ti.
|
11 |
It outperforms Jukebox representation on GTZAN (genre classification) and GiantSteps (key classification) datasets.
|
12 |
Larger models trained with more data are on the way~
|