Classifier_with_external_sets_04

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

• Loss: 0.2741
• Accuracy: 0.9193

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 26
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9983	289	0.3540	0.8893
0.5256	2.0	579	0.3874	0.8673
0.5256	2.9983	868	0.3275	0.8991
0.378	4.0	1158	0.3244	0.9028
0.378	4.9983	1447	0.4013	0.8312
0.4029	6.0	1737	0.4052	0.8428
0.3932	6.9983	2026	0.3667	0.8801
0.3932	8.0	2316	0.3972	0.8385
0.3972	8.9983	2605	0.3983	0.8648
0.3972	10.0	2895	0.3805	0.8587
0.3734	10.9983	3184	0.3735	0.8746
0.3734	12.0	3474	0.3256	0.8893
0.3752	12.9983	3763	0.2800	0.9101
0.3169	14.0	4053	0.3071	0.8979
0.3169	14.9983	4342	0.3083	0.9052
0.312	16.0	4632	0.2894	0.9168
0.312	16.9983	4921	0.3725	0.8624
0.3162	18.0	5211	0.3163	0.8979
0.3185	18.9983	5500	0.3030	0.8991
0.3185	20.0	5790	0.3045	0.8997
0.2951	20.9983	6079	0.2944	0.9076
0.2951	22.0	6369	0.2693	0.9199
0.2916	22.9983	6658	0.2711	0.9187
0.2916	24.0	6948	0.2651	0.9211
0.2593	24.9983	7237	0.2696	0.9193
0.2646	25.9551	7514	0.2741	0.9193

Framework versions

• Transformers 4.40.0
• Pytorch 2.2.2+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1