scenario-NON-KD-PR-COPY-CDF-CL-D2_data-cl-cardiff_cl_only44

This model is a fine-tuned version of microsoft/mdeberta-v3-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 5.8690
• Accuracy: 0.4468
• F1: 0.4461

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: 32
• seed: 44
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	1.09	250	1.3853	0.4421	0.4355
0.8929	2.17	500	1.5079	0.4414	0.4346
0.8929	3.26	750	1.8059	0.4483	0.4460
0.5584	4.35	1000	2.0344	0.4414	0.4386
0.5584	5.43	1250	2.2842	0.4282	0.4210
0.2846	6.52	1500	2.5446	0.4576	0.4503
0.2846	7.61	1750	3.3811	0.4259	0.4202
0.1506	8.7	2000	3.4956	0.4460	0.4445
0.1506	9.78	2250	3.9632	0.4352	0.4289
0.0896	10.87	2500	3.8448	0.4529	0.4524
0.0896	11.96	2750	3.6707	0.4483	0.4484
0.0633	13.04	3000	4.1145	0.4375	0.4377
0.0633	14.13	3250	4.2206	0.4383	0.4372
0.052	15.22	3500	4.6658	0.4653	0.4649
0.052	16.3	3750	4.7310	0.4468	0.4458
0.0298	17.39	4000	5.2055	0.4398	0.4387
0.0298	18.48	4250	4.8530	0.4360	0.4321
0.0303	19.57	4500	5.0298	0.4398	0.4395
0.0303	20.65	4750	5.2417	0.4491	0.4494
0.0127	21.74	5000	5.6795	0.4437	0.4422
0.0127	22.83	5250	5.3289	0.4437	0.4434
0.0124	23.91	5500	5.5139	0.4460	0.4455
0.0124	25.0	5750	5.7711	0.4321	0.4261
0.0092	26.09	6000	5.7323	0.4429	0.4414
0.0092	27.17	6250	5.7387	0.4483	0.4481
0.0054	28.26	6500	5.8869	0.4444	0.4439
0.0054	29.35	6750	5.8690	0.4468	0.4461

Framework versions

• Transformers 4.33.3
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.13.3