exceptions_exp2_swap_0.7_last_to_drop_5039

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.5624
• Accuracy: 0.3685

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: 0.0006
• train_batch_size: 16
• eval_batch_size: 16
• seed: 5039
• gradient_accumulation_steps: 5
• total_train_batch_size: 80
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.98) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 50.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
4.8112	0.2915	1000	4.7410	0.2563
4.3438	0.5830	2000	4.2855	0.2988
4.1468	0.8745	3000	4.1017	0.3149
4.0037	1.1659	4000	3.9990	0.3243
3.9334	1.4574	5000	3.9216	0.3309
3.8726	1.7489	6000	3.8626	0.3361
3.7435	2.0402	7000	3.8201	0.3405
3.7597	2.3317	8000	3.7867	0.3434
3.742	2.6233	9000	3.7586	0.3463
3.7362	2.9148	10000	3.7326	0.3481
3.6481	3.2061	11000	3.7197	0.3505
3.6483	3.4976	12000	3.7008	0.3525
3.6519	3.7891	13000	3.6839	0.3538
3.5549	4.0805	14000	3.6782	0.3553
3.573	4.3720	15000	3.6657	0.3562
3.5669	4.6635	16000	3.6552	0.3573
3.5748	4.9550	17000	3.6404	0.3586
3.4992	5.2463	18000	3.6417	0.3591
3.5202	5.5378	19000	3.6309	0.3601
3.5416	5.8293	20000	3.6185	0.3611
3.4494	6.1207	21000	3.6253	0.3615
3.4666	6.4122	22000	3.6162	0.3620
3.4982	6.7037	23000	3.6074	0.3627
3.491	6.9952	24000	3.5966	0.3635
3.4383	7.2866	25000	3.6028	0.3634
3.4471	7.5781	26000	3.5959	0.3641
3.459	7.8696	27000	3.5882	0.3650
3.3868	8.1609	28000	3.5967	0.3646
3.4131	8.4524	29000	3.5888	0.3653
3.4209	8.7439	30000	3.5793	0.3661
3.3226	9.0353	31000	3.5848	0.3663
3.3821	9.3268	32000	3.5840	0.3663
3.387	9.6183	33000	3.5783	0.3669
3.4285	9.9098	34000	3.5683	0.3678
3.3364	10.2011	35000	3.5799	0.3669
3.3779	10.4927	36000	3.5727	0.3676
3.3912	10.7842	37000	3.5632	0.3681
3.2965	11.0755	38000	3.5753	0.3682
3.3433	11.3670	39000	3.5694	0.3681
3.3688	11.6585	40000	3.5624	0.3685
3.3747	11.9500	41000	3.5556	0.3694
3.3082	12.2414	42000	3.5720	0.3690
3.3454	12.5329	43000	3.5629	0.3693
3.3472	12.8244	44000	3.5550	0.3697
3.274	13.1157	45000	3.5676	0.3692
3.3233	13.4072	46000	3.5669	0.3693
3.3412	13.6988	47000	3.5537	0.3700
3.3467	13.9903	48000	3.5483	0.3706
3.2855	14.2816	49000	3.5637	0.3699
3.3095	14.5731	50000	3.5562	0.3703
3.3091	14.8646	51000	3.5488	0.3705
3.2558	15.1560	52000	3.5630	0.3699
3.2926	15.4475	53000	3.5574	0.3705
3.3081	15.7390	54000	3.5482	0.3710
3.2065	16.0303	55000	3.5619	0.3705
3.2603	16.3218	56000	3.5620	0.3706
3.2886	16.6133	57000	3.5528	0.3711
3.2956	16.9049	58000	3.5434	0.3716
3.2221	17.1962	59000	3.5621	0.3706
3.2614	17.4877	60000	3.5547	0.3713
3.2888	17.7792	61000	3.5491	0.3718
3.1966	18.0705	62000	3.5610	0.3710
3.2353	18.3621	63000	3.5548	0.3714
3.2483	18.6536	64000	3.5492	0.3717
3.2875	18.9451	65000	3.5427	0.3720
3.203	19.2364	66000	3.5579	0.3715
3.2456	19.5279	67000	3.5545	0.3718
3.2715	19.8194	68000	3.5433	0.3727
3.1963	20.1108	69000	3.5555	0.3719
3.2248	20.4023	70000	3.5534	0.3721
3.2486	20.6938	71000	3.5476	0.3723
3.2651	20.9853	72000	3.5375	0.3730
3.2031	21.2766	73000	3.5569	0.3719
3.2173	21.5682	74000	3.5453	0.3725
3.2354	21.8597	75000	3.5416	0.3728
3.1755	22.1510	76000	3.5609	0.3717
3.2026	22.4425	77000	3.5525	0.3723
3.2214	22.7340	78000	3.5460	0.3728
3.1373	23.0254	79000	3.5615	0.3722
3.1901	23.3169	80000	3.5565	0.3724
3.2152	23.6084	81000	3.5498	0.3728
3.2334	23.8999	82000	3.5425	0.3732
3.1535	24.1912	83000	3.5609	0.3724
3.1899	24.4827	84000	3.5494	0.3729
3.2195	24.7743	85000	3.5458	0.3733
3.1314	25.0656	86000	3.5594	0.3726
3.1665	25.3571	87000	3.5585	0.3728
3.196	25.6486	88000	3.5520	0.3732
3.2065	25.9401	89000	3.5428	0.3736
3.1425	26.2315	90000	3.5591	0.3729
3.179	26.5230	91000	3.5549	0.3733
3.1973	26.8145	92000	3.5469	0.3737

Framework versions

• Transformers 4.55.2
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4