learning_rate: 0.0006
train_batch_size: 16
eval_batch_size: 16
seed: 1032
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
4.8248	0.2917	1000	4.7620	0.2540
4.3406	0.5834	2000	4.2890	0.2984
4.1489	0.8750	3000	4.1041	0.3145
3.9987	1.1665	4000	3.9986	0.3235
3.9326	1.4582	5000	3.9220	0.3305
3.8882	1.7499	6000	3.8644	0.3361
3.7546	2.0414	7000	3.8234	0.3396
3.764	2.3331	8000	3.7900	0.3430
3.7419	2.6248	9000	3.7618	0.3459
3.7225	2.9165	10000	3.7343	0.3484
3.6388	3.2080	11000	3.7232	0.3503
3.6638	3.4996	12000	3.7044	0.3518
3.6478	3.7913	13000	3.6864	0.3537
3.5579	4.0828	14000	3.6806	0.3546
3.5815	4.3745	15000	3.6711	0.3559
3.5842	4.6662	16000	3.6555	0.3571
3.583	4.9579	17000	3.6432	0.3583
3.5165	5.2494	18000	3.6462	0.3588
3.5252	5.5411	19000	3.6340	0.3597
3.5507	5.8327	20000	3.6241	0.3605
3.455	6.1243	21000	3.6269	0.3610
3.4813	6.4159	22000	3.6178	0.3621
3.5001	6.7076	23000	3.6084	0.3628
3.5018	6.9993	24000	3.6007	0.3634
3.4475	7.2908	25000	3.6083	0.3633
3.4491	7.5825	26000	3.5996	0.3640
3.4656	7.8742	27000	3.5922	0.3645
3.4013	8.1657	28000	3.5995	0.3646
3.4293	8.4574	29000	3.5913	0.3653
3.4323	8.7490	30000	3.5854	0.3655
3.3299	9.0405	31000	3.5907	0.3657
3.3788	9.3322	32000	3.5892	0.3659
3.4058	9.6239	33000	3.5806	0.3664
3.4188	9.9156	34000	3.5720	0.3670
3.3383	10.2071	35000	3.5819	0.3668
3.3669	10.4988	36000	3.5775	0.3675
3.3929	10.7905	37000	3.5692	0.3677
3.3083	11.0820	38000	3.5793	0.3676
3.3325	11.3736	39000	3.5746	0.3677
3.3675	11.6653	40000	3.5648	0.3685
3.3735	11.9570	41000	3.5574	0.3689
3.3232	12.2485	42000	3.5710	0.3686
3.3335	12.5402	43000	3.5629	0.3691
3.3466	12.8319	44000	3.5589	0.3695
3.2702	13.1234	45000	3.5689	0.3688
3.3164	13.4151	46000	3.5643	0.3691
3.3378	13.7067	47000	3.5567	0.3697
3.3462	13.9984	48000	3.5483	0.3703
3.2838	14.2899	49000	3.5651	0.3695
3.3036	14.5816	50000	3.5569	0.3702
3.3254	14.8733	51000	3.5556	0.3701
3.2589	15.1648	52000	3.5678	0.3697
3.2853	15.4565	53000	3.5603	0.3703
3.3027	15.7482	54000	3.5535	0.3703
3.2245	16.0397	55000	3.5661	0.3702
3.2718	16.3313	56000	3.5584	0.3705
3.2702	16.6230	57000	3.5498	0.3710
3.3126	16.9147	58000	3.5469	0.3714
3.2443	17.2062	59000	3.5668	0.3704
3.2604	17.4979	60000	3.5541	0.3712
3.2926	17.7896	61000	3.5467	0.3717
3.1954	18.0811	62000	3.5618	0.3709
3.2459	18.3728	63000	3.5565	0.3714
3.2651	18.6644	64000	3.5510	0.3715
3.2855	18.9561	65000	3.5430	0.3721
3.2179	19.2476	66000	3.5620	0.3711
3.2502	19.5393	67000	3.5504	0.3718
3.2587	19.8310	68000	3.5435	0.3722
3.1811	20.1225	69000	3.5604	0.3715
3.227	20.4142	70000	3.5571	0.3714
3.2613	20.7059	71000	3.5472	0.3721
3.2561	20.9975	72000	3.5449	0.3723
3.2074	21.2891	73000	3.5570	0.3720
3.2265	21.5807	74000	3.5502	0.3725
3.2463	21.8724	75000	3.5453	0.3728
3.1738	22.1639	76000	3.5614	0.3718
3.2158	22.4556	77000	3.5527	0.3724
3.2276	22.7473	78000	3.5468	0.3728
3.1475	23.0388	79000	3.5580	0.3722
3.1835	23.3305	80000	3.5616	0.3719
3.2186	23.6222	81000	3.5483	0.3728
3.2286	23.9138	82000	3.5435	0.3732
3.1577	24.2053	83000	3.5560	0.3727
3.1893	24.4970	84000	3.5525	0.3728
3.2137	24.7887	85000	3.5451	0.3730

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_cost_to_drop_1032

exceptions_exp2_swap_0.7_cost_to_drop_1032

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

craa/
exceptions_exp2_swap_0.7_cost_to_drop_1032