exceptions_exp2_swap_0.3_last_to_carry_3591

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

• Loss: 3.5612
• Accuracy: 0.3691

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: 3591
• 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
4.8399	0.2915	1000	4.7682	0.2526
4.3417	0.5830	2000	4.2962	0.2976
4.1439	0.8745	3000	4.1005	0.3142
4.0005	1.1659	4000	3.9900	0.3249
3.9362	1.4574	5000	3.9157	0.3314
3.8926	1.7488	6000	3.8588	0.3366
3.7472	2.0402	7000	3.8146	0.3409
3.7684	2.3317	8000	3.7864	0.3439
3.7338	2.6232	9000	3.7552	0.3465
3.7199	2.9147	10000	3.7305	0.3490
3.6476	3.2061	11000	3.7164	0.3511
3.6576	3.4976	12000	3.7000	0.3524
3.6399	3.7891	13000	3.6801	0.3543
3.5544	4.0805	14000	3.6764	0.3553
3.5757	4.3719	15000	3.6623	0.3564
3.5834	4.6634	16000	3.6514	0.3577
3.5868	4.9549	17000	3.6386	0.3588
3.5078	5.2463	18000	3.6392	0.3595
3.5205	5.5378	19000	3.6287	0.3602
3.5423	5.8293	20000	3.6174	0.3612
3.436	6.1207	21000	3.6212	0.3615
3.4739	6.4122	22000	3.6156	0.3621
3.496	6.7037	23000	3.6022	0.3631
3.5004	6.9952	24000	3.5970	0.3637
3.4317	7.2865	25000	3.6048	0.3635
3.4552	7.5780	26000	3.5947	0.3644
3.4603	7.8695	27000	3.5853	0.3648
3.3768	8.1609	28000	3.5968	0.3649
3.4119	8.4524	29000	3.5881	0.3653
3.427	8.7439	30000	3.5792	0.3663
3.3273	9.0353	31000	3.5856	0.3662
3.3851	9.3268	32000	3.5841	0.3666
3.4055	9.6183	33000	3.5750	0.3670
3.4297	9.9098	34000	3.5666	0.3675
3.3332	10.2011	35000	3.5812	0.3672
3.3678	10.4926	36000	3.5713	0.3676
3.3949	10.7841	37000	3.5624	0.3683
3.2867	11.0755	38000	3.5737	0.3683
3.3385	11.3670	39000	3.5688	0.3683
3.368	11.6585	40000	3.5612	0.3691
3.3772	11.9500	41000	3.5559	0.3692
3.32	12.2414	42000	3.5702	0.3685
3.3344	12.5329	43000	3.5591	0.3695
3.3551	12.8243	44000	3.5531	0.3700
3.2865	13.1157	45000	3.5688	0.3692
3.3038	13.4072	46000	3.5627	0.3695
3.3348	13.6987	47000	3.5542	0.3701
3.3473	13.9902	48000	3.5451	0.3704
3.2939	14.2816	49000	3.5636	0.3700
3.309	14.5731	50000	3.5579	0.3704
3.3242	14.8646	51000	3.5464	0.3707
3.2465	15.1559	52000	3.5644	0.3704
3.2815	15.4474	53000	3.5569	0.3706
3.2962	15.7389	54000	3.5495	0.3712
3.2177	16.0303	55000	3.5609	0.3706
3.2562	16.3218	56000	3.5605	0.3710
3.2811	16.6133	57000	3.5487	0.3713
3.3011	16.9048	58000	3.5415	0.3720
3.2295	17.1962	59000	3.5593	0.3708
3.2583	17.4877	60000	3.5512	0.3712
3.2796	17.7792	61000	3.5451	0.3717
3.2001	18.0705	62000	3.5596	0.3711
3.2379	18.3620	63000	3.5548	0.3715
3.2711	18.6535	64000	3.5500	0.3717
3.2752	18.9450	65000	3.5378	0.3725
3.2176	19.2364	66000	3.5540	0.3716
3.2513	19.5279	67000	3.5475	0.3720
3.2672	19.8194	68000	3.5396	0.3726
3.1738	20.1108	69000	3.5589	0.3717
3.2262	20.4023	70000	3.5498	0.3725
3.2411	20.6938	71000	3.5441	0.3729
3.2383	20.9853	72000	3.5354	0.3733
3.1997	21.2766	73000	3.5551	0.3720
3.2377	21.5681	74000	3.5509	0.3725
3.2425	21.8596	75000	3.5385	0.3732
3.1718	22.1510	76000	3.5542	0.3726
3.206	22.4425	77000	3.5503	0.3727
3.2306	22.7340	78000	3.5421	0.3732
3.1304	23.0254	79000	3.5559	0.3726
3.1792	23.3169	80000	3.5586	0.3724
3.2102	23.6083	81000	3.5432	0.3732
3.2296	23.8998	82000	3.5398	0.3736
3.1541	24.1912	83000	3.5567	0.3728
3.1828	24.4827	84000	3.5520	0.3732
3.212	24.7742	85000	3.5426	0.3736
3.1323	25.0656	86000	3.5610	0.3727
3.168	25.3571	87000	3.5539	0.3731
3.1908	25.6486	88000	3.5477	0.3732
3.2038	25.9401	89000	3.5389	0.3741
3.1401	26.2314	90000	3.5552	0.3731
3.1825	26.5229	91000	3.5477	0.3737
3.1995	26.8144	92000	3.5389	0.3742

Framework versions

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