exceptions_exp2_swap_last_to_push_2128

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

• Loss: 3.5625
• Accuracy: 0.3688

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: 2128
• 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
4.8414	0.2915	1000	4.7598	0.2536
4.3522	0.5830	2000	4.2918	0.2981
4.1577	0.8744	3000	4.1090	0.3140
4.0044	1.1659	4000	3.9980	0.3238
3.9382	1.4573	5000	3.9235	0.3305
3.8916	1.7488	6000	3.8664	0.3355
3.7463	2.0402	7000	3.8246	0.3401
3.76	2.3317	8000	3.7934	0.3431
3.7499	2.6232	9000	3.7619	0.3459
3.7326	2.9147	10000	3.7369	0.3483
3.6296	3.2061	11000	3.7239	0.3500
3.656	3.4976	12000	3.7059	0.3517
3.6405	3.7890	13000	3.6874	0.3537
3.5577	4.0804	14000	3.6793	0.3549
3.5639	4.3719	15000	3.6670	0.3561
3.5875	4.6634	16000	3.6542	0.3573
3.5855	4.9549	17000	3.6395	0.3585
3.5175	5.2463	18000	3.6420	0.3590
3.5229	5.5378	19000	3.6326	0.3599
3.536	5.8293	20000	3.6210	0.3609
3.4536	6.1207	21000	3.6254	0.3613
3.4721	6.4121	22000	3.6183	0.3619
3.492	6.7036	23000	3.6074	0.3628
3.514	6.9951	24000	3.5988	0.3634
3.447	7.2865	25000	3.6066	0.3634
3.4663	7.5780	26000	3.5981	0.3639
3.4629	7.8695	27000	3.5878	0.3652
3.4138	8.1609	28000	3.5994	0.3645
3.4229	8.4524	29000	3.5896	0.3652
3.4336	8.7438	30000	3.5803	0.3660
3.3422	9.0353	31000	3.5879	0.3662
3.3935	9.3267	32000	3.5842	0.3661
3.3983	9.6182	33000	3.5757	0.3666
3.42	9.9097	34000	3.5689	0.3675
3.3412	10.2011	35000	3.5815	0.3671
3.3878	10.4926	36000	3.5739	0.3673
3.3928	10.7841	37000	3.5676	0.3682
3.2972	11.0755	38000	3.5754	0.3681
3.3434	11.3670	39000	3.5705	0.3683
3.3577	11.6584	40000	3.5625	0.3688
3.3797	11.9499	41000	3.5562	0.3692
3.296	12.2413	42000	3.5719	0.3688
3.3427	12.5328	43000	3.5653	0.3688
3.343	12.8243	44000	3.5552	0.3697
3.2751	13.1157	45000	3.5670	0.3694
3.3112	13.4072	46000	3.5626	0.3695
3.3342	13.6987	47000	3.5544	0.3703
3.3426	13.9901	48000	3.5471	0.3706
3.2837	14.2816	49000	3.5615	0.3701
3.3151	14.5730	50000	3.5549	0.3704
3.3292	14.8645	51000	3.5483	0.3709
3.2577	15.1559	52000	3.5618	0.3702
3.2903	15.4474	53000	3.5573	0.3706
3.3151	15.7389	54000	3.5504	0.3709
3.206	16.0303	55000	3.5577	0.3709
3.2612	16.3218	56000	3.5585	0.3707
3.2838	16.6133	57000	3.5499	0.3713
3.2954	16.9047	58000	3.5414	0.3718
3.2272	17.1962	59000	3.5593	0.3713
3.2686	17.4876	60000	3.5552	0.3712
3.2794	17.7791	61000	3.5425	0.3721
3.2079	18.0705	62000	3.5554	0.3715
3.249	18.3620	63000	3.5515	0.3717
3.277	18.6535	64000	3.5443	0.3723
3.2772	18.9450	65000	3.5406	0.3724
3.2091	19.2364	66000	3.5568	0.3718
3.2416	19.5279	67000	3.5483	0.3724
3.2611	19.8193	68000	3.5399	0.3728
3.1814	20.1108	69000	3.5568	0.3722
3.2103	20.4022	70000	3.5525	0.3721
3.2462	20.6937	71000	3.5419	0.3728
3.2769	20.9852	72000	3.5369	0.3733
3.2038	21.2766	73000	3.5526	0.3723
3.2176	21.5681	74000	3.5475	0.3727
3.2234	21.8596	75000	3.5388	0.3730
3.1661	22.1510	76000	3.5585	0.3723
3.2196	22.4425	77000	3.5509	0.3727
3.2282	22.7339	78000	3.5441	0.3731
3.1333	23.0254	79000	3.5540	0.3726
3.1819	23.3168	80000	3.5537	0.3727
3.2117	23.6083	81000	3.5426	0.3732
3.2366	23.8998	82000	3.5382	0.3736
3.1715	24.1912	83000	3.5596	0.3728
3.1758	24.4827	84000	3.5493	0.3732
3.2168	24.7742	85000	3.5404	0.3736
3.1238	25.0656	86000	3.5571	0.3731
3.1759	25.3571	87000	3.5471	0.3733
3.1958	25.6485	88000	3.5457	0.3734
3.2297	25.9400	89000	3.5405	0.3738
3.1469	26.2314	90000	3.5577	0.3731
3.1871	26.5229	91000	3.5456	0.3736
3.1779	26.8144	92000	3.5392	0.3741

Framework versions

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