scenario-NON-KD-SCR-D2_data-cl-massive_all_1_166

This model is a fine-tuned version of microsoft/mdeberta-v3-base on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 10.7320
• Accuracy: 0.1825
• F1: 0.1740

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: 5e-05
• train_batch_size: 32
• eval_batch_size: 64
• seed: 66
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.0633	0.5558	5000	3.8974	0.1789	0.1652
0.5994	1.1116	10000	4.3223	0.1915	0.1845
0.5992	1.6674	15000	4.3654	0.1928	0.1844
0.3702	2.2232	20000	5.2309	0.1920	0.1795
0.3825	2.7790	25000	4.7960	0.1883	0.1815
0.2343	3.3348	30000	6.5335	0.1861	0.1784
0.2594	3.8906	35000	6.3502	0.1851	0.1802
0.1633	4.4464	40000	6.9355	0.1810	0.1694
0.167	5.0022	45000	7.5505	0.1828	0.1771
0.1137	5.5580	50000	8.4792	0.1828	0.1751
0.0661	6.1138	55000	9.2692	0.1830	0.1726
0.0731	6.6696	60000	9.2159	0.1804	0.1676
0.0437	7.2254	65000	9.9592	0.1789	0.1721
0.0504	7.7812	70000	10.3607	0.1805	0.1712
0.0317	8.3370	75000	10.6761	0.1804	0.1723
0.0264	8.8928	80000	10.8029	0.1823	0.1737
0.0161	9.4486	85000	10.7320	0.1825	0.1740

Framework versions

• Transformers 4.44.2
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.19.1