Update README.md
Browse files
README.md
CHANGED
|
@@ -14,4 +14,9 @@ tags:
|
|
| 14 |
|
| 15 |
This model is part of the [GrammarCorrector](https://github.com/akhmat-s/GrammarCorrector) tool.
|
| 16 |
|
| 17 |
-
"[FlanT5 from scratch for the grammar correction tool](https://medium.com/@akhmat-s/flant5-from-scratch-for-the-grammar-correction-tool-deadba9a6778)" article about how this
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
This model is part of the [GrammarCorrector](https://github.com/akhmat-s/GrammarCorrector) tool.
|
| 16 |
|
| 17 |
+
"[FlanT5 from scratch for the grammar correction tool](https://medium.com/@akhmat-s/flant5-from-scratch-for-the-grammar-correction-tool-deadba9a6778)" article about how this models was trained:
|
| 18 |
+
>FlanT5 was trained using [JFLEG](https://arxiv.org/abs/1702.04066) dataset. The primary objective of the experiment was to develop a highly effective tool using relatively small models, minimal datasets, and constrained computational resources.
|
| 19 |
+
>
|
| 20 |
+
>To accomplish this goal, we implemented two key strategies:
|
| 21 |
+
>- [Perplexity-Based Data](https://arxiv.org/abs/2405.20541) Pruning With Small Reference Models.
|
| 22 |
+
>- A simple sampling and voting method for [multiple LLM agents](https://arxiv.org/abs/2402.05120). model was trained.
|