FLANG-BERT_bert-base-uncased

This model is a fine-tuned version of SALT-NLP/FLANG-BERT on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7138
• Accuracy: 0.8643
• F1: 0.8645
• Precision: 0.8681
• Recall: 0.8643

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.8614	1.0	91	0.8043	0.6443	0.6279	0.6398	0.6443
0.5386	2.0	182	0.4807	0.8112	0.8113	0.8150	0.8112
0.3422	3.0	273	0.4452	0.8300	0.8304	0.8351	0.8300
0.2617	4.0	364	0.5424	0.8190	0.8177	0.8259	0.8190
0.164	5.0	455	0.5162	0.8424	0.8414	0.8424	0.8424
0.1278	6.0	546	0.5737	0.8440	0.8439	0.8440	0.8440
0.0599	7.0	637	0.6869	0.8268	0.8236	0.8311	0.8268
0.1184	8.0	728	0.5331	0.8471	0.8475	0.8493	0.8471
0.1126	9.0	819	0.6979	0.8237	0.8221	0.8332	0.8237
0.0737	10.0	910	0.7481	0.8362	0.8362	0.8381	0.8362
0.1425	11.0	1001	0.7602	0.8315	0.8308	0.8331	0.8315
0.0666	12.0	1092	0.6645	0.8612	0.8612	0.8615	0.8612
0.0523	13.0	1183	0.7138	0.8643	0.8645	0.8681	0.8643
0.0168	14.0	1274	0.7317	0.8534	0.8525	0.8527	0.8534
0.0336	15.0	1365	0.8575	0.8456	0.8454	0.8553	0.8456
0.0424	16.0	1456	0.9331	0.8409	0.8386	0.8423	0.8409
0.0188	17.0	1547	0.7885	0.8596	0.8595	0.8599	0.8596
0.0032	18.0	1638	0.8774	0.8596	0.8584	0.8592	0.8596

Framework versions

• Transformers 4.37.0
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.1