Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,23 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: cc-by-4.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
library_name: transformers
|
| 6 |
+
pipeline_tag: text-classification
|
| 7 |
+
tags:
|
| 8 |
+
- code
|
| 9 |
+
metrics:
|
| 10 |
+
- accuracy
|
| 11 |
+
- f1
|
| 12 |
+
---
|
| 13 |
+
# CodeBERT-SO
|
| 14 |
+
Repository for CodeBERT, fine-tuned on Stack Overflow snippets with respect to NL-PL pairs of 6 languages (Python, Java, JavaScript, PHP, Ruby, Go).
|
| 15 |
+
## Training Objective
|
| 16 |
+
This model is initialized with [CodeBERT-base](https://huggingface.co/microsoft/codebert-base) and trained to classify whether a user will drop out given their posts and code snippets.
|
| 17 |
+
## Training Regime
|
| 18 |
+
Training was done across 8 epochs with a batch size of 8, learning rate of 1e-5, epsilon (weight update denominator) of 1e-8.
|
| 19 |
+
A random 20% sample of the entire dataset was used as the validation set.
|
| 20 |
+
## Performance
|
| 21 |
+
* Final validation accuracy: 0.822
|
| 22 |
+
* Final validation F1: 0.809
|
| 23 |
+
* Final validation loss: 0.5
|