Add upstream attribution notice
Browse files
NOTICE.md
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Notices
|
| 2 |
+
|
| 3 |
+
## This repository
|
| 4 |
+
|
| 5 |
+
Core ML conversion of the laya-multilingual decision model, by Fluid Inference.
|
| 6 |
+
Licensed under the Apache License, Version 2.0; see `LICENSE`.
|
| 7 |
+
|
| 8 |
+
## Upstream model and code
|
| 9 |
+
|
| 10 |
+
The converted weights and the reference implementation are the work of
|
| 11 |
+
Convai Innovations and contributors, used under the Apache License, Version 2.0.
|
| 12 |
+
|
| 13 |
+
- Weights: [`convaiinnovations/laya`](https://huggingface.co/convaiinnovations/laya),
|
| 14 |
+
subfolder `multilingual`, revision `1c5edc17a7acd8701df6fc341c0d179f1c62c982`.
|
| 15 |
+
- Reference implementation: [`NandhaKishorM/laya`](https://github.com/NandhaKishorM/laya).
|
| 16 |
+
|
| 17 |
+
The encoder is [`jhu-clsp/mmBERT-base`](https://huggingface.co/jhu-clsp/mmBERT-base),
|
| 18 |
+
and `tokenizer.json` carries its vocabulary, redistributed under that model's terms.
|
| 19 |
+
|
| 20 |
+
No weights were retrained. The conversion changes tensor precision and fixes input
|
| 21 |
+
shapes; the parameters are the upstream ones. This is an independent conversion and
|
| 22 |
+
is not an official Convai Innovations release.
|