Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
This model is used to detect **Offensive Content** in **Kannada Code-Mixed language**. The mono in the name refers to the monolingual setting, where the model is trained using only Kannada(pure and code-mixed) data. The weights are initialized from pretrained XLM-Roberta-Base and pretrained using Masked Language Modelling on the target dataset before fine-tuning using Cross-Entropy Loss.
|
| 2 |
|
| 3 |
This model is the best of multiple trained for **EACL 2021 Shared Task on Offensive Language Identification in Dravidian Languages**. Genetic-Algorithm based ensembled test predictions got the second-highest weighted F1 score at the leaderboard (Weighted F1 score on hold out test set: This model - 0.73, Ensemble - 0.74)
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: kn
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
---
|
| 5 |
This model is used to detect **Offensive Content** in **Kannada Code-Mixed language**. The mono in the name refers to the monolingual setting, where the model is trained using only Kannada(pure and code-mixed) data. The weights are initialized from pretrained XLM-Roberta-Base and pretrained using Masked Language Modelling on the target dataset before fine-tuning using Cross-Entropy Loss.
|
| 6 |
|
| 7 |
This model is the best of multiple trained for **EACL 2021 Shared Task on Offensive Language Identification in Dravidian Languages**. Genetic-Algorithm based ensembled test predictions got the second-highest weighted F1 score at the leaderboard (Weighted F1 score on hold out test set: This model - 0.73, Ensemble - 0.74)
|