Update README.md
Browse files
README.md
CHANGED
|
@@ -6,7 +6,9 @@ language:
|
|
| 6 |
- en
|
| 7 |
pipeline_tag: automatic-speech-recognition
|
| 8 |
---
|
| 9 |
-
|
|
|
|
|
|
|
| 10 |
<div align="center" style="line-height: 1;">
|
| 11 |
<a href="https://github.com/augustgw/early-exit-transformer" target="_blank" style="margin: 2px;">
|
| 12 |
<img alt="GitHub" src="https://img.shields.io/badge/GitHub-Splitformer-181717?logo=github&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
|
|
@@ -19,11 +21,6 @@ pipeline_tag: automatic-speech-recognition
|
|
| 19 |
</a>
|
| 20 |
</div>
|
| 21 |
|
| 22 |
-
---
|
| 23 |
-
|
| 24 |
-
# SplitFormer
|
| 25 |
-
|
| 26 |
-
## Overview
|
| 27 |
Splitformer is a Conformer-based ASR model (36.7M parameters) trained from scratch on 1000 hours of the LibriSpeech dataset with an early‐exit objective.
|
| 28 |
|
| 29 |
This architecture introduces parallel downsampling layers before the first and last exits to improve performance with minimal extra overhead, while retaining inference speed.
|
|
|
|
| 6 |
- en
|
| 7 |
pipeline_tag: automatic-speech-recognition
|
| 8 |
---
|
| 9 |
+
|
| 10 |
+
# SplitFormer
|
| 11 |
+
|
| 12 |
<div align="center" style="line-height: 1;">
|
| 13 |
<a href="https://github.com/augustgw/early-exit-transformer" target="_blank" style="margin: 2px;">
|
| 14 |
<img alt="GitHub" src="https://img.shields.io/badge/GitHub-Splitformer-181717?logo=github&logoColor=white" style="display: inline-block; vertical-align: middle;"/>
|
|
|
|
| 21 |
</a>
|
| 22 |
</div>
|
| 23 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
Splitformer is a Conformer-based ASR model (36.7M parameters) trained from scratch on 1000 hours of the LibriSpeech dataset with an early‐exit objective.
|
| 25 |
|
| 26 |
This architecture introduces parallel downsampling layers before the first and last exits to improve performance with minimal extra overhead, while retaining inference speed.
|