exceptions_exp2_swap_0.3_last_to_drop_3591

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

• Loss: 3.5611
• Accuracy: 0.3692

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
4.8307	0.2915	1000	4.7528	0.2547
4.3399	0.5830	2000	4.2888	0.2988
4.1412	0.8745	3000	4.0976	0.3150
3.999	1.1659	4000	3.9911	0.3247
3.9366	1.4574	5000	3.9156	0.3315
3.8932	1.7488	6000	3.8589	0.3366
3.7472	2.0402	7000	3.8156	0.3406
3.7682	2.3317	8000	3.7874	0.3440
3.7343	2.6232	9000	3.7559	0.3465
3.7197	2.9147	10000	3.7309	0.3487
3.6492	3.2061	11000	3.7177	0.3511
3.6586	3.4976	12000	3.6991	0.3525
3.6414	3.7891	13000	3.6815	0.3543
3.5561	4.0805	14000	3.6759	0.3554
3.577	4.3719	15000	3.6633	0.3563
3.5849	4.6634	16000	3.6510	0.3577
3.587	4.9549	17000	3.6385	0.3591
3.509	5.2463	18000	3.6400	0.3596
3.5213	5.5378	19000	3.6289	0.3604
3.5433	5.8293	20000	3.6175	0.3614
3.4371	6.1207	21000	3.6213	0.3615
3.4752	6.4122	22000	3.6150	0.3622
3.4955	6.7037	23000	3.6015	0.3633
3.5007	6.9952	24000	3.5953	0.3639
3.4315	7.2865	25000	3.6035	0.3638
3.4545	7.5780	26000	3.5948	0.3646
3.4606	7.8695	27000	3.5853	0.3654
3.3771	8.1609	28000	3.5933	0.3651
3.412	8.4524	29000	3.5879	0.3655
3.4258	8.7439	30000	3.5777	0.3663
3.3274	9.0353	31000	3.5839	0.3664
3.3857	9.3268	32000	3.5825	0.3668
3.4067	9.6183	33000	3.5739	0.3672
3.4288	9.9098	34000	3.5648	0.3679
3.3344	10.2011	35000	3.5775	0.3675
3.3684	10.4926	36000	3.5698	0.3681
3.394	10.7841	37000	3.5625	0.3685
3.2872	11.0755	38000	3.5743	0.3684
3.3392	11.3670	39000	3.5675	0.3683
3.3675	11.6585	40000	3.5611	0.3692
3.377	11.9500	41000	3.5557	0.3693
3.3208	12.2414	42000	3.5674	0.3687
3.3352	12.5329	43000	3.5571	0.3696
3.3566	12.8243	44000	3.5539	0.3703
3.2877	13.1157	45000	3.5664	0.3696
3.304	13.4072	46000	3.5602	0.3700
3.3343	13.6987	47000	3.5514	0.3703
3.348	13.9902	48000	3.5433	0.3707
3.2943	14.2816	49000	3.5629	0.3701
3.309	14.5731	50000	3.5542	0.3708
3.3244	14.8646	51000	3.5442	0.3712
3.2472	15.1559	52000	3.5590	0.3708
3.2813	15.4474	53000	3.5533	0.3710
3.2968	15.7389	54000	3.5465	0.3716
3.2181	16.0303	55000	3.5576	0.3709
3.2577	16.3218	56000	3.5567	0.3710
3.2813	16.6133	57000	3.5463	0.3716
3.3009	16.9048	58000	3.5392	0.3722
3.2308	17.1962	59000	3.5570	0.3713
3.2594	17.4877	60000	3.5497	0.3717
3.279	17.7792	61000	3.5440	0.3720
3.2003	18.0705	62000	3.5573	0.3714
3.2396	18.3620	63000	3.5517	0.3719
3.2702	18.6535	64000	3.5449	0.3723
3.2764	18.9450	65000	3.5358	0.3728
3.217	19.2364	66000	3.5490	0.3719
3.2504	19.5279	67000	3.5482	0.3722
3.2675	19.8194	68000	3.5371	0.3730
3.1747	20.1108	69000	3.5570	0.3720
3.2261	20.4023	70000	3.5480	0.3729
3.2419	20.6938	71000	3.5406	0.3730
3.2373	20.9853	72000	3.5357	0.3732
3.2006	21.2766	73000	3.5510	0.3724
3.2397	21.5681	74000	3.5452	0.3728
3.2436	21.8596	75000	3.5327	0.3735
3.1737	22.1510	76000	3.5524	0.3729
3.2065	22.4425	77000	3.5469	0.3728
3.2299	22.7340	78000	3.5368	0.3734
3.1314	23.0254	79000	3.5550	0.3730
3.1791	23.3169	80000	3.5532	0.3728
3.2121	23.6083	81000	3.5387	0.3733
3.2298	23.8998	82000	3.5360	0.3739
3.1542	24.1912	83000	3.5532	0.3730
3.1837	24.4827	84000	3.5471	0.3735
3.212	24.7742	85000	3.5388	0.3738
3.1334	25.0656	86000	3.5558	0.3730
3.168	25.3571	87000	3.5501	0.3733
3.1905	25.6486	88000	3.5455	0.3734
3.2034	25.9401	89000	3.5347	0.3745
3.1402	26.2314	90000	3.5523	0.3736
3.1817	26.5229	91000	3.5460	0.3740
3.2001	26.8144	92000	3.5371	0.3744
3.1347	27.1058	93000	3.5538	0.3734
3.1525	27.3973	94000	3.5473	0.3737
3.1895	27.6888	95000	3.5404	0.3740

Framework versions

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