dit-small_tobacco3482_kd_CEKD_t5.0_a0.9

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: 2.4735
• Accuracy: 0.19
• Brier Loss: 0.8651
• Nll: 6.3618
• F1 Micro: 0.19
• F1 Macro: 0.0641
• Ece: 0.2456
• Aurc: 0.7331

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	2.6674	0.06	0.9042	9.2824	0.06	0.0114	0.1749	0.9042
No log	1.96	6	2.5911	0.18	0.8886	6.4746	0.18	0.0305	0.2317	0.8026
No log	2.96	9	2.5252	0.18	0.8764	7.5079	0.18	0.0305	0.2390	0.8141
No log	3.96	12	2.5235	0.185	0.8777	6.9489	0.185	0.0488	0.2553	0.7838
No log	4.96	15	2.5223	0.185	0.8754	6.8606	0.185	0.0488	0.2572	0.7773
No log	5.96	18	2.5213	0.185	0.8732	5.9794	0.185	0.0488	0.2384	0.7684
No log	6.96	21	2.5203	0.185	0.8723	5.9244	0.185	0.0488	0.2406	0.7603
No log	7.96	24	2.5149	0.185	0.8713	5.9034	0.185	0.0488	0.2484	0.7560
No log	8.96	27	2.5064	0.185	0.8701	5.9325	0.185	0.0488	0.2525	0.7529
No log	9.96	30	2.5014	0.185	0.8695	6.7123	0.185	0.0488	0.2399	0.7528
No log	10.96	33	2.4977	0.185	0.8693	6.7598	0.185	0.0488	0.2487	0.7511
No log	11.96	36	2.4944	0.185	0.8692	6.8130	0.185	0.0488	0.2488	0.7476
No log	12.96	39	2.4908	0.185	0.8688	6.7610	0.185	0.0488	0.2488	0.7452
No log	13.96	42	2.4867	0.185	0.8680	6.6686	0.185	0.0488	0.2484	0.7428
No log	14.96	45	2.4830	0.185	0.8673	6.6283	0.185	0.0488	0.2426	0.7431
No log	15.96	48	2.4805	0.185	0.8668	6.4857	0.185	0.0488	0.2385	0.7410
No log	16.96	51	2.4794	0.185	0.8666	6.4425	0.185	0.0488	0.2459	0.7385
No log	17.96	54	2.4795	0.185	0.8664	6.0769	0.185	0.0488	0.2406	0.7352
No log	18.96	57	2.4793	0.185	0.8664	6.1000	0.185	0.0488	0.2402	0.7355
No log	19.96	60	2.4774	0.185	0.8660	6.3802	0.185	0.0488	0.2506	0.7346
No log	20.96	63	2.4762	0.185	0.8657	6.4330	0.185	0.0488	0.2550	0.7344
No log	21.96	66	2.4750	0.185	0.8654	6.3721	0.185	0.0488	0.2513	0.7333
No log	22.96	69	2.4741	0.19	0.8652	6.3676	0.19	0.0641	0.2453	0.7332
No log	23.96	72	2.4738	0.19	0.8652	6.3645	0.19	0.0641	0.2455	0.7331
No log	24.96	75	2.4735	0.19	0.8651	6.3618	0.19	0.0641	0.2456	0.7331

Framework versions

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