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	Accuracy	Validation Loss
4.8465	0.2915	1000	0.2536	4.7656
4.3527	0.5831	2000	0.2978	4.3010
4.1587	0.8746	3000	0.3141	4.1073
4.0102	1.1662	4000	0.3242	4.0016
3.9355	1.4577	5000	0.3307	3.9212
3.8747	1.7493	6000	0.3357	3.8646
3.733	2.0408	7000	0.3402	3.8186
3.762	2.3324	8000	0.3435	3.7873
3.7482	2.6239	9000	0.3460	3.7596
3.7228	2.9155	10000	0.3484	3.7306
3.6399	3.2070	11000	0.3506	3.7207
3.6553	3.4985	12000	0.3522	3.7024
3.6382	3.7901	13000	0.3540	3.6841
3.5475	4.0816	14000	0.3554	3.6782
3.5792	4.3732	15000	0.3561	3.6679
3.5881	4.6647	16000	0.3575	3.6521
3.5807	4.9563	17000	0.3584	3.6384
3.5165	5.2478	18000	0.3591	3.6417
3.5147	5.5394	19000	0.3601	3.6306
3.5401	5.8309	20000	0.3609	3.6210
3.4533	6.1224	21000	0.3614	3.6238
3.4765	6.4140	22000	0.3619	3.6173
3.492	6.7055	23000	0.3630	3.6073
3.4955	6.9971	24000	0.3635	3.5966
3.4402	7.2886	25000	0.3634	3.6059
3.4633	7.5802	26000	0.3641	3.5958
3.4632	7.8717	27000	0.3649	3.5856
3.3815	8.1633	28000	0.3645	3.5969
3.424	8.4548	29000	0.3654	3.5898
3.43	8.7464	30000	0.3659	3.5823
3.3299	9.0379	31000	0.3660	3.5859
3.3741	9.3294	32000	0.3661	3.5825
3.402	9.6210	33000	0.3668	3.5770
3.4142	9.9125	34000	0.3674	3.5687
3.341	10.2041	35000	0.3671	3.5819
3.3621	10.4956	36000	0.3673	3.5752
3.3982	10.7872	37000	0.3679	3.5666
3.309	11.0787	38000	0.3678	3.5780
3.3408	11.3703	39000	0.3680	3.5726
3.3394	11.6618	40000	0.3686	3.5622
3.3773	11.9534	41000	0.3692	3.5532
3.297	12.2449	42000	0.3686	3.5664
3.3416	12.5364	43000	0.3692	3.5612
3.3651	12.8280	44000	0.3694	3.5560
3.2801	13.1195	45000	0.3691	3.5695
3.3202	13.4111	46000	0.3696	3.5630
3.3334	13.7026	47000	0.3697	3.5554
3.3351	13.9942	48000	0.3705	3.5472
3.2906	14.2857	49000	0.3699	3.5628
3.3026	14.5773	50000	0.3699	3.5579
3.3281	14.8688	51000	0.3710	3.5470
3.2502	15.1603	52000	0.3701	3.5615
3.2882	15.4519	53000	0.3708	3.5550
3.3117	15.7434	54000	0.3711	3.5493
3.2029	16.0350	55000	0.3703	3.5622
3.2678	16.3265	56000	0.3710	3.5591
3.2947	16.6181	57000	0.3713	3.5477
3.3098	16.9096	58000	0.3719	3.5414
3.2327	17.2012	59000	0.3711	3.5589
3.2576	17.4927	60000	0.3714	3.5522
3.2814	17.7843	61000	0.3719	3.5424
3.2031	18.0758	62000	0.3717	3.5551
3.2527	18.3673	63000	0.3716	3.5570
3.2609	18.6589	64000	0.3720	3.5470
3.2827	18.9504	65000	0.3723	3.5409
3.2155	19.2420	66000	0.3713	3.5597
3.2566	19.5335	67000	0.3723	3.5504
3.267	19.8251	68000	0.3725	3.5413
3.1863	20.1166	69000	0.3716	3.5572
3.2187	20.4082	70000	0.3722	3.5539
3.2443	20.6997	71000	0.3726	3.5455
3.2638	20.9913	72000	0.3729	3.5361
3.2133	21.2828	73000	0.3722	3.5517
3.2314	21.5743	74000	0.3725	3.5497
3.2541	21.8659	75000	0.3729	3.5422
3.1691	22.1574	76000	0.3720	3.5563
3.2007	22.4490	77000	0.3726	3.5541
3.2393	22.7405	78000	0.3730	3.5424
3.1345	23.0321	79000	0.3728	3.5545
3.1809	23.3236	80000	0.3727	3.5525
3.1754	23.6152	81000	3.5585	0.3723
3.2106	23.9067	82000	3.5483	0.3731
3.1535	24.1983	83000	3.5610	0.3725
3.2012	24.4898	84000	3.5533	0.3729
3.2107	24.7813	85000	3.5419	0.3735
3.1402	25.0729	86000	3.5566	0.3728
3.1835	25.3644	87000	3.5531	0.3731
3.197	25.6560	88000	3.5481	0.3733
3.2256	25.9475	89000	3.5389	0.3738
3.1436	26.2391	90000	3.5587	0.3729
3.1725	26.5306	91000	3.5479	0.3734
3.2054	26.8222	92000	3.5433	0.3738
3.1216	27.1137	93000	3.5582	0.3729
3.1637	27.4052	94000	3.5531	0.3733
3.1803	27.6968	95000	3.5426	0.3737
3.212	27.9883	96000	3.5385	0.3738
3.1403	28.2799	97000	3.5584	0.3732
3.1622	28.5714	98000	3.5465	0.3737
3.1771	28.8630	99000	3.5416	0.3740
3.1284	29.1545	100000	3.5594	0.3732

Framework versions

Transformers 4.55.2
Pytorch 2.8.0+cu128
Datasets 4.0.0
Tokenizers 0.21.4

craa/exceptions_exp2_swap_0.7_resemble_to_drop_2128

exceptions_exp2_swap_0.7_resemble_to_drop_2128

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

craa/
exceptions_exp2_swap_0.7_resemble_to_drop_2128