|
--- |
|
license: apache-2.0 |
|
--- |
|
### Deprem NER Training Results |
|
|
|
``` |
|
training_args = TrainingArguments( |
|
output_dir="./output", |
|
evaluation_strategy="epoch", |
|
per_device_train_batch_size=32, |
|
per_device_eval_batch_size=32, |
|
weight_decay=0.01, |
|
report_to=None, |
|
num_train_epochs=4 |
|
) |
|
``` |
|
|
|
Threshold: 0.1 |
|
|
|
``` |
|
precision recall f1-score support |
|
|
|
Alakasiz 0.92 0.87 0.89 734 |
|
Barinma 0.87 0.79 0.83 207 |
|
Elektronik 0.72 0.73 0.73 130 |
|
Giysi 0.84 0.66 0.74 94 |
|
Kurtarma 0.84 0.80 0.82 362 |
|
Lojistik 0.75 0.51 0.61 112 |
|
Saglik 0.79 0.80 0.79 108 |
|
Su 0.63 0.47 0.54 78 |
|
Yagma 0.75 0.58 0.65 31 |
|
Yemek 0.80 0.77 0.79 117 |
|
|
|
micro avg 0.85 0.78 0.81 1973 |
|
macro avg 0.79 0.70 0.74 1973 |
|
weighted avg 0.84 0.78 0.81 1973 |
|
samples avg 0.84 0.82 0.82 1973 |
|
``` |
