|
|
--- |
|
|
language: |
|
|
- am |
|
|
size_categories: |
|
|
- 1K<n<10K |
|
|
task_categories: |
|
|
- text-classification |
|
|
dataset_info: |
|
|
features: |
|
|
- name: clean_tweet |
|
|
dtype: string |
|
|
- name: label |
|
|
dtype: |
|
|
class_label: |
|
|
names: |
|
|
'0': negative |
|
|
'1': positive |
|
|
splits: |
|
|
- name: train |
|
|
num_bytes: 468510 |
|
|
num_examples: 2223 |
|
|
- name: dev |
|
|
num_bytes: 56319 |
|
|
num_examples: 279 |
|
|
- name: test |
|
|
num_bytes: 58731 |
|
|
num_examples: 279 |
|
|
download_size: 338974 |
|
|
dataset_size: 583560 |
|
|
configs: |
|
|
- config_name: default |
|
|
data_files: |
|
|
- split: train |
|
|
path: data/train-* |
|
|
- split: dev |
|
|
path: data/dev-* |
|
|
- split: test |
|
|
path: data/test-* |
|
|
--- |
|
|
|
|
|
# Amharic Sentiment Dataset |
|
|
This dataset contains `2781` cleaned Amharic Tweets, labeled as having either `positive` or `negative` sentiment. This dataset can be used to train a sentiment classification model. |
|
|
|
|
|
### Dataset Source |
|
|
|
|
|
https://github.com/liyaSileshi/amharic-sentiment-analysis/blob/main/data_preprocess/train.csv |
|
|
|
|
|
### Finetuned Models |
|
|
|
|
|
The following models were finetuned using this dataset. The reported precision, recall, and f1 metrics are macro averages. |
|
|
|
|
|
|Model|Size (# params)| Accuracy | Precision | Recall | F1 | |
|
|
| --- | ------------- | -------- | --------- | ------ | -- | |
|
|
|bert-medium-amharic|40.5M|0.83|0.83|0.82|0.83| |
|
|
|bert-small-amharic|27.8M|0.83|0.83|0.82|0.83| |
|
|
|bert-mini-amharic|10.7M|0.81|0.81|0.81|0.81| |
|
|
|bert-tiny-amharic|4.18M|0.79|0.79|0.79|0.79| |
|
|
|xlm-roberta-base|279M|0.83|0.83|0.83|0.83| |
|
|
|am-roberta|443M|0.82|0.83|0.82|0.82| |
|
|
|
|
|
#### Code |
|
|
In this repository, you can find notebooks for finetuning each of the above models using this dataset |
|
|
|
|
|
- https://github.com/rasyosef/amharic-sentiment-classification |
|
|
|
