dit-tiny_tobacco3482_kd_CEKD_t2.5_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: 3.2510
• Accuracy: 0.18
• Brier Loss: 0.8767
• Nll: 6.8039
• F1 Micro: 0.18
• F1 Macro: 0.0306
• Ece: 0.2513
• Aurc: 0.8508

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	3.4586	0.145	0.8999	10.1587	0.145	0.0253	0.2221	0.8467
No log	1.96	6	3.4232	0.145	0.8946	10.5824	0.145	0.0253	0.2242	0.8475
No log	2.96	9	3.3704	0.16	0.8867	8.6135	0.16	0.0503	0.2171	0.8440
No log	3.96	12	3.3273	0.155	0.8807	6.5471	0.155	0.0274	0.2248	0.8831
No log	4.96	15	3.3006	0.155	0.8779	6.8045	0.155	0.0271	0.2331	0.8918
No log	5.96	18	3.2856	0.16	0.8773	8.2046	0.16	0.0329	0.2361	0.8956
No log	6.96	21	3.2758	0.18	0.8774	8.0738	0.18	0.0308	0.2561	0.8544
No log	7.96	24	3.2688	0.18	0.8778	7.1046	0.18	0.0308	0.2647	0.8524
No log	8.96	27	3.2630	0.18	0.8778	6.9910	0.18	0.0306	0.2591	0.8530
No log	9.96	30	3.2597	0.18	0.8778	6.9680	0.18	0.0306	0.2736	0.8538
No log	10.96	33	3.2573	0.18	0.8776	6.9547	0.18	0.0306	0.2698	0.8536
No log	11.96	36	3.2557	0.18	0.8775	6.9491	0.18	0.0306	0.2653	0.8533
No log	12.96	39	3.2546	0.18	0.8773	6.8987	0.18	0.0306	0.2606	0.8526
No log	13.96	42	3.2536	0.18	0.8771	6.8204	0.18	0.0306	0.2601	0.8523
No log	14.96	45	3.2528	0.18	0.8771	6.8141	0.18	0.0306	0.2521	0.8519
No log	15.96	48	3.2522	0.18	0.8769	6.8074	0.18	0.0306	0.2606	0.8517
No log	16.96	51	3.2519	0.18	0.8769	6.8077	0.18	0.0306	0.2607	0.8515
No log	17.96	54	3.2520	0.18	0.8769	6.8050	0.18	0.0306	0.2561	0.8510
No log	18.96	57	3.2520	0.18	0.8769	6.8057	0.18	0.0306	0.2519	0.8509
No log	19.96	60	3.2515	0.18	0.8768	6.8046	0.18	0.0306	0.2556	0.8507
No log	20.96	63	3.2514	0.18	0.8768	6.8048	0.18	0.0306	0.2515	0.8506
No log	21.96	66	3.2512	0.18	0.8767	6.8048	0.18	0.0306	0.2556	0.8508
No log	22.96	69	3.2510	0.18	0.8767	6.8045	0.18	0.0306	0.2513	0.8509
No log	23.96	72	3.2510	0.18	0.8767	6.8043	0.18	0.0306	0.2513	0.8508
No log	24.96	75	3.2510	0.18	0.8767	6.8039	0.18	0.0306	0.2513	0.8508

Framework versions

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