Update README.md
Browse files
README.md
CHANGED
|
@@ -1,6 +1,15 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
Model results on SentEval Benchmark:
|
| 5 |
```shell
|
| 6 |
+-------+-------+-------+-------+-------+--------------+-----------------+--------+
|
|
@@ -8,4 +17,9 @@ Model results on SentEval Benchmark:
|
|
| 8 |
+-------+-------+-------+-------+-------+--------------+-----------------+--------+
|
| 9 |
| 71.71 | 83.09 | 75.46 | 83.13 | 80.22 | 79.70 | 73.62 | 78.13 |
|
| 10 |
+-------+-------+-------+-------+-------+--------------+-----------------+--------+
|
| 11 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
| 4 |
+
# mutual information Contrastive Sentence Embedding (**miCSE**):
|
| 5 |
+
|
| 6 |
+
Language model trained for sentence similarity using mutual information-based Contrastive learning.
|
| 7 |
+
|
| 8 |
+
Training **miCSE** imposes alignment between the attention pattern of different views during contrastive learning. Learning sentence embeddings with **miCSE** entails enforcing the syntactic consistency across augmented views for every single sentence, making contrastive self-supervised learning more sample efficient.
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
# Benchmark
|
| 12 |
+
|
| 13 |
Model results on SentEval Benchmark:
|
| 14 |
```shell
|
| 15 |
+-------+-------+-------+-------+-------+--------------+-----------------+--------+
|
|
|
|
| 17 |
+-------+-------+-------+-------+-------+--------------+-----------------+--------+
|
| 18 |
| 71.71 | 83.09 | 75.46 | 83.13 | 80.22 | 79.70 | 73.62 | 78.13 |
|
| 19 |
+-------+-------+-------+-------+-------+--------------+-----------------+--------+
|
| 20 |
+
```
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
#### Authors:
|
| 24 |
+
- [Tassilo Klein](https://tjklein.github.io/)
|
| 25 |
+
- [Moin Nabi](https://moinnabi.github.io/)
|