dit-tiny_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: 2.5147
• Accuracy: 0.18
• Brier Loss: 0.8746
• Nll: 6.7241
• F1 Micro: 0.18
• F1 Macro: 0.0306
• Ece: 0.2451
• Aurc: 0.8494

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	2.6571	0.145	0.8999	10.1542	0.145	0.0253	0.2220	0.8466
No log	1.96	6	2.6281	0.145	0.8947	10.5635	0.145	0.0253	0.2236	0.8461
No log	2.96	9	2.5865	0.14	0.8870	8.5822	0.14	0.0433	0.2063	0.8040
No log	3.96	12	2.5552	0.19	0.8811	6.5445	0.19	0.0552	0.2421	0.8576
No log	4.96	15	2.5387	0.155	0.8782	7.1184	0.155	0.0277	0.2280	0.8892
No log	5.96	18	2.5317	0.18	0.8774	8.7285	0.18	0.0319	0.2392	0.8538
No log	6.96	21	2.5274	0.18	0.8770	8.2533	0.18	0.0306	0.2476	0.8514
No log	7.96	24	2.5238	0.18	0.8767	6.9903	0.18	0.0306	0.2465	0.8523
No log	8.96	27	2.5205	0.18	0.8762	6.9049	0.18	0.0306	0.2473	0.8528
No log	9.96	30	2.5189	0.18	0.8758	6.8830	0.18	0.0306	0.2515	0.8526
No log	10.96	33	2.5180	0.18	0.8756	6.8133	0.18	0.0306	0.2469	0.8522
No log	11.96	36	2.5175	0.18	0.8754	6.7422	0.18	0.0306	0.2500	0.8519
No log	12.96	39	2.5173	0.18	0.8753	6.5762	0.18	0.0306	0.2533	0.8515
No log	13.96	42	2.5168	0.18	0.8751	6.5666	0.18	0.0306	0.2528	0.8516
No log	14.96	45	2.5164	0.18	0.8750	6.7246	0.18	0.0306	0.2532	0.8512
No log	15.96	48	2.5160	0.18	0.8750	6.7221	0.18	0.0306	0.2456	0.8507
No log	16.96	51	2.5157	0.18	0.8749	6.7242	0.18	0.0306	0.2457	0.8507
No log	17.96	54	2.5158	0.18	0.8749	6.7241	0.18	0.0306	0.2417	0.8503
No log	18.96	57	2.5157	0.18	0.8749	6.7259	0.18	0.0306	0.2455	0.8503
No log	19.96	60	2.5153	0.18	0.8748	6.7248	0.18	0.0306	0.2452	0.8495
No log	20.96	63	2.5151	0.18	0.8748	6.7250	0.18	0.0306	0.2414	0.8494
No log	21.96	66	2.5149	0.18	0.8747	6.7250	0.18	0.0306	0.2452	0.8495
No log	22.96	69	2.5147	0.18	0.8747	6.7247	0.18	0.0306	0.2451	0.8495
No log	23.96	72	2.5147	0.18	0.8747	6.7246	0.18	0.0306	0.2451	0.8495
No log	24.96	75	2.5147	0.18	0.8746	6.7241	0.18	0.0306	0.2451	0.8494

Framework versions

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