dit-small_tobacco3482_kd_CEKD_t5.0_a0.7

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

• Loss: 3.1347
• Accuracy: 0.185
• Brier Loss: 0.8666
• Nll: 5.9997
• F1 Micro: 0.185
• F1 Macro: 0.0488
• Ece: 0.2480
• Aurc: 0.7353

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	0.96	3	3.3695	0.06	0.9042	9.1505	0.06	0.0114	0.1750	0.9033
No log	1.96	6	3.2847	0.18	0.8890	7.1646	0.18	0.0305	0.2263	0.8027
No log	2.96	9	3.2039	0.18	0.8773	8.6118	0.18	0.0305	0.2478	0.8186
No log	3.96	12	3.1950	0.18	0.8806	7.4891	0.18	0.0305	0.2514	0.8131
No log	4.96	15	3.1951	0.185	0.8795	6.7125	0.185	0.0488	0.2555	0.7835
No log	5.96	18	3.1931	0.185	0.8766	5.2600	0.185	0.0488	0.2526	0.7702
No log	6.96	21	3.1876	0.185	0.8741	5.6453	0.185	0.0488	0.2372	0.7672
No log	7.96	24	3.1800	0.185	0.8726	5.9473	0.185	0.0488	0.2412	0.7644
No log	8.96	27	3.1712	0.185	0.8712	5.9421	0.185	0.0488	0.2491	0.7615
No log	9.96	30	3.1656	0.185	0.8704	6.6276	0.185	0.0488	0.2516	0.7602
No log	10.96	33	3.1623	0.185	0.8704	6.8796	0.185	0.0488	0.2487	0.7598
No log	11.96	36	3.1601	0.185	0.8708	7.1352	0.185	0.0488	0.2451	0.7559
No log	12.96	39	3.1573	0.185	0.8706	7.0151	0.185	0.0488	0.2492	0.7531
No log	13.96	42	3.1531	0.185	0.8699	6.7912	0.185	0.0488	0.2450	0.7484
No log	14.96	45	3.1485	0.185	0.8693	6.6578	0.185	0.0488	0.2513	0.7491
No log	15.96	48	3.1449	0.185	0.8685	6.1407	0.185	0.0488	0.2596	0.7463
No log	16.96	51	3.1428	0.185	0.8681	5.9160	0.185	0.0488	0.2548	0.7432
No log	17.96	54	3.1421	0.185	0.8678	5.8419	0.185	0.0488	0.2449	0.7401
No log	18.96	57	3.1413	0.185	0.8677	5.7417	0.185	0.0488	0.2606	0.7382
No log	19.96	60	3.1391	0.185	0.8673	5.7824	0.185	0.0488	0.2432	0.7365
No log	20.96	63	3.1378	0.185	0.8671	5.9509	0.185	0.0488	0.2598	0.7368
No log	21.96	66	3.1364	0.185	0.8668	6.0164	0.185	0.0488	0.2477	0.7361
No log	22.96	69	3.1355	0.185	0.8667	6.0109	0.185	0.0488	0.2437	0.7352
No log	23.96	72	3.1350	0.185	0.8666	6.0029	0.185	0.0488	0.2438	0.7351
No log	24.96	75	3.1347	0.185	0.8666	5.9997	0.185	0.0488	0.2480	0.7353

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2