albert-base-ours-run-2

This model is a fine-tuned version of albert-base-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.2462
• Accuracy: 0.695
• Precision: 0.6550
• Recall: 0.6529
• F1: 0.6539

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.999	1.0	200	0.9155	0.615	0.5590	0.5590	0.5524
0.7736	2.0	400	0.8488	0.6	0.5639	0.5689	0.5256
0.5836	3.0	600	0.8760	0.67	0.6259	0.6158	0.6191
0.4153	4.0	800	1.0050	0.675	0.6356	0.6212	0.5974
0.3188	5.0	1000	1.2033	0.655	0.6254	0.5977	0.5991
0.2335	6.0	1200	1.3407	0.625	0.5955	0.6039	0.5937
0.1752	7.0	1400	1.4246	0.72	0.6846	0.6815	0.6820
0.1056	8.0	1600	1.9654	0.69	0.6589	0.6251	0.6311
0.0696	9.0	1800	1.9376	0.715	0.6908	0.6632	0.6627
0.0352	10.0	2000	1.9970	0.72	0.6880	0.6784	0.6817
0.0227	11.0	2200	2.1449	0.705	0.6901	0.6641	0.6679
0.0199	12.0	2400	2.2213	0.72	0.6891	0.6685	0.6749
0.0077	13.0	2600	2.1500	0.69	0.6729	0.6704	0.6647
0.0067	14.0	2800	2.1780	0.69	0.6632	0.6651	0.6621
0.0034	15.0	3000	2.1759	0.71	0.6800	0.6786	0.6788
0.0013	16.0	3200	2.2139	0.71	0.6760	0.6721	0.6735
0.0005	17.0	3400	2.2282	0.7	0.6606	0.6593	0.6599
0.0003	18.0	3600	2.2257	0.7	0.6606	0.6593	0.6599
0.0003	19.0	3800	2.2492	0.695	0.6550	0.6529	0.6539
0.0002	20.0	4000	2.2462	0.695	0.6550	0.6529	0.6539

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.0+cu116
• Tokenizers 0.13.2