dit-tiny_tobacco3482_kd_CEKD_t2.5_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: 2.5379
• Accuracy: 0.18
• Brier Loss: 0.8746
• Nll: 6.7389
• F1 Micro: 0.18
• F1 Macro: 0.0306
• Ece: 0.2460
• Aurc: 0.8496

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	2.6891	0.145	0.8999	10.1550	0.145	0.0253	0.2220	0.8466
No log	1.96	6	2.6592	0.145	0.8947	10.5706	0.145	0.0253	0.2238	0.8463
No log	2.96	9	2.6158	0.14	0.8869	8.5528	0.14	0.0422	0.2066	0.8175
No log	3.96	12	2.5827	0.175	0.8810	6.5464	0.175	0.0467	0.2385	0.8661
No log	4.96	15	2.5647	0.155	0.8781	6.8570	0.155	0.0274	0.2316	0.8886
No log	5.96	18	2.5566	0.19	0.8772	8.4283	0.19	0.0413	0.2460	0.8532
No log	6.96	21	2.5515	0.18	0.8769	7.6865	0.18	0.0308	0.2480	0.8517
No log	7.96	24	2.5475	0.18	0.8767	6.9727	0.18	0.0306	0.2469	0.8521
No log	8.96	27	2.5438	0.18	0.8762	6.9080	0.18	0.0306	0.2438	0.8525
No log	9.96	30	2.5420	0.18	0.8758	6.8906	0.18	0.0306	0.2521	0.8528
No log	10.96	33	2.5410	0.18	0.8755	6.8317	0.18	0.0306	0.2516	0.8524
No log	11.96	36	2.5404	0.18	0.8753	6.7606	0.18	0.0306	0.2469	0.8516
No log	12.96	39	2.5401	0.18	0.8752	6.7444	0.18	0.0306	0.2425	0.8516
No log	13.96	42	2.5397	0.18	0.8751	6.7397	0.18	0.0306	0.2498	0.8514
No log	14.96	45	2.5393	0.18	0.8750	6.7390	0.18	0.0306	0.2579	0.8511
No log	15.96	48	2.5389	0.18	0.8749	6.7366	0.18	0.0306	0.2463	0.8513
No log	16.96	51	2.5387	0.18	0.8749	6.7390	0.18	0.0306	0.2465	0.8510
No log	17.96	54	2.5389	0.18	0.8749	6.7382	0.18	0.0306	0.2425	0.8505
No log	18.96	57	2.5389	0.18	0.8749	6.7397	0.18	0.0306	0.2463	0.8504
No log	19.96	60	2.5384	0.18	0.8748	6.7391	0.18	0.0306	0.2421	0.8495
No log	20.96	63	2.5383	0.18	0.8747	6.7396	0.18	0.0306	0.2422	0.8500
No log	21.96	66	2.5380	0.18	0.8747	6.7399	0.18	0.0306	0.2460	0.8496
No log	22.96	69	2.5379	0.18	0.8746	6.7395	0.18	0.0306	0.2460	0.8497
No log	23.96	72	2.5379	0.18	0.8746	6.7393	0.18	0.0306	0.2460	0.8497
No log	24.96	75	2.5379	0.18	0.8746	6.7389	0.18	0.0306	0.2460	0.8496

Framework versions

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