Javier Huertas
commited on
Commit
·
9103237
1
Parent(s):
92e6d25
include config
Browse files
README.md
CHANGED
|
@@ -1,3 +1,33 @@
|
|
| 1 |
-
---
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pipeline_tag: text-classification
|
| 3 |
+
tags:
|
| 4 |
+
- natural-language-inference
|
| 5 |
+
- misogyny
|
| 6 |
+
language: en
|
| 7 |
+
license: apache-2.0
|
| 8 |
+
widget:
|
| 9 |
+
- text: "Las mascarillas causan hipoxia. Wearing masks is harmful to human health"
|
| 10 |
+
example_title: "Natural Language Inference"
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# bertweet-base-multi-mami
|
| 14 |
+
This is a finetuned XLM-RoBERTA model for natural language inference. It has been trained with a massive ammount of data following the ANLI pipeline training. We include data from:
|
| 15 |
+
- [mnli](https://cims.nyu.edu/~sbowman/multinli/) {train, dev and test}
|
| 16 |
+
- [snli](https://nlp.stanford.edu/projects/snli/) {train, dev and test}
|
| 17 |
+
- [xnli](https://github.com/facebookresearch/XNLI) {train, dev and test}
|
| 18 |
+
- [fever](https://fever.ai/resources.html) {train, dev and test}
|
| 19 |
+
- [anli](https://github.com/facebookresearch/anli) {train}
|
| 20 |
+
|
| 21 |
+
The model is validated on ANLI training sets, including R1, R2 and R3. The following results can be expected on the testing splits.
|
| 22 |
+
|Split|Accuracy|
|
| 23 |
+
|-|-|
|
| 24 |
+
|R1|0.6610|
|
| 25 |
+
|R2|0.4990|
|
| 26 |
+
|R3|0.4425|
|
| 27 |
+
|
| 28 |
+
# Multilabels
|
| 29 |
+
label2id={
|
| 30 |
+
"contradiction": 0,
|
| 31 |
+
"entailment": 1,
|
| 32 |
+
"neutral": 2,
|
| 33 |
+
},
|