albert_model

This model is a fine-tuned version of albert-base-v2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6560
• Accuracy: 0.9070
• F1: 0.8852
• Recall: 0.9122

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Recall
No log	1.0	167	0.3571	0.8351	0.8142	0.9198
No log	2.0	334	0.2670	0.8891	0.8683	0.9313
0.3358	3.0	501	0.2643	0.9115	0.8885	0.8969
0.3358	4.0	668	0.3804	0.9130	0.8910	0.9046
0.3358	5.0	835	0.4376	0.9070	0.8848	0.9084
0.1007	6.0	1002	0.4957	0.9100	0.8859	0.8893
0.1007	7.0	1169	0.6375	0.8801	0.8601	0.9389
0.1007	8.0	1336	0.5978	0.8996	0.8780	0.9198
0.012	9.0	1503	0.6101	0.9025	0.8816	0.9237
0.012	10.0	1670	0.6209	0.9085	0.8847	0.8931
0.012	11.0	1837	0.6485	0.9010	0.8787	0.9122
0.0007	12.0	2004	0.6480	0.9070	0.8852	0.9122
0.0007	13.0	2171	0.6527	0.9055	0.8835	0.9122
0.0007	14.0	2338	0.6557	0.9055	0.8835	0.9122
0.0002	15.0	2505	0.6560	0.9070	0.8852	0.9122

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3