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
4.8322	0.2915	1000	4.7517	0.2546
4.337	0.5831	2000	4.2911	0.2983
4.1649	0.8746	3000	4.1033	0.3144
4.0042	1.1662	4000	3.9973	0.3237
3.9465	1.4577	5000	3.9266	0.3303
3.8922	1.7493	6000	3.8672	0.3356
3.7474	2.0408	7000	3.8251	0.3398
3.7632	2.3324	8000	3.7972	0.3428
3.7465	2.6239	9000	3.7645	0.3458
3.733	2.9155	10000	3.7380	0.3480
3.6534	3.2070	11000	3.7235	0.3500
3.6578	3.4985	12000	3.7055	0.3520
3.6495	3.7901	13000	3.6886	0.3532
3.5494	4.0816	14000	3.6828	0.3545
3.5758	4.3732	15000	3.6684	0.3558
3.5983	4.6647	16000	3.6560	0.3569
3.6045	4.9563	17000	3.6429	0.3583
3.5129	5.2478	18000	3.6454	0.3587
3.5234	5.5394	19000	3.6358	0.3598
3.5342	5.8309	20000	3.6220	0.3607
3.4452	6.1224	21000	3.6246	0.3610
3.4895	6.4140	22000	3.6183	0.3620
3.4913	6.7055	23000	3.6103	0.3626
3.4937	6.9971	24000	3.6003	0.3634
3.4264	7.2886	25000	3.6071	0.3635
3.4557	7.5802	26000	3.6000	0.3639
3.4678	7.8717	27000	3.5903	0.3648
3.3794	8.1633	28000	3.6007	0.3646
3.4321	8.4548	29000	3.5920	0.3650
3.4377	8.7464	30000	3.5846	0.3654
3.3411	9.0379	31000	3.5890	0.3658
3.3879	9.3294	32000	3.5857	0.3660
3.4095	9.6210	33000	3.5792	0.3667
3.4237	9.9125	34000	3.5699	0.3673
3.3486	10.2041	35000	3.5815	0.3671
3.3843	10.4956	36000	3.5722	0.3674
3.3956	10.7872	37000	3.5673	0.3681
3.3032	11.0787	38000	3.5800	0.3677
3.3429	11.3703	39000	3.5753	0.3678
3.3597	11.6618	40000	3.5662	0.3685
3.3828	11.9534	41000	3.5586	0.3690
3.3189	12.2449	42000	3.5699	0.3683
3.3526	12.5364	43000	3.5650	0.3690
3.3583	12.8280	44000	3.5575	0.3695
3.2758	13.1195	45000	3.5714	0.3689
3.3109	13.4111	46000	3.5671	0.3694
3.3436	13.7026	47000	3.5575	0.3699
3.364	13.9942	48000	3.5506	0.3702
3.2776	14.2857	49000	3.5658	0.3694
3.3105	14.5773	50000	3.5611	0.3701
3.3296	14.8688	51000	3.5497	0.3705
3.2426	15.1603	52000	3.5653	0.3703
3.3094	15.4519	53000	3.5594	0.3704
3.3121	15.7434	54000	3.5535	0.3710
3.2062	16.0350	55000	3.5619	0.3705
3.2574	16.3265	56000	3.5601	0.3706
3.2957	16.6181	57000	3.5561	0.3710
3.3022	16.9096	58000	3.5457	0.3718
3.2422	17.2012	59000	3.5625	0.3709
3.2727	17.4927	60000	3.5530	0.3715
3.2841	17.7843	61000	3.5449	0.3718
3.2102	18.0758	62000	3.5590	0.3714
3.2347	18.3673	63000	3.5562	0.3714
3.2737	18.6589	64000	3.5493	0.3720
3.2915	18.9504	65000	3.5385	0.3725
3.2183	19.2420	66000	3.5586	0.3714
3.254	19.5335	67000	3.5513	0.3719
3.2793	19.8251	68000	3.5438	0.3724
3.1916	20.1166	69000	3.5584	0.3716
3.219	20.4082	70000	3.5566	0.3717
3.2467	20.6997	71000	3.5445	0.3725
3.266	20.9913	72000	3.5381	0.3728
3.2023	21.2828	73000	3.5587	0.3720
3.2291	21.5743	74000	3.5498	0.3723
3.2598	21.8659	75000	3.5407	0.3730
3.1798	22.1574	76000	3.5571	0.3721
3.2057	22.4490	77000	3.5546	0.3725
3.2431	22.7405	78000	3.5465	0.3727
3.1456	23.0321	79000	3.5572	0.3727
3.2018	23.3236	80000	3.5522	0.3729
3.2088	23.6152	81000	3.5490	0.3731
3.2195	23.9067	82000	3.5386	0.3732
3.16	24.1983	83000	3.5546	0.3728
3.1915	24.4898	84000	3.5520	0.3730
3.2081	24.7813	85000	3.5437	0.3733
3.1358	25.0729	86000	3.5630	0.3726
3.1746	25.3644	87000	3.5521	0.3732
3.1914	25.6560	88000	3.5463	0.3733
3.2154	25.9475	89000	3.5404	0.3740
3.1581	26.2391	90000	3.5589	0.3730
3.1753	26.5306	91000	3.5498	0.3734
3.2032	26.8222	92000	3.5433	0.3739

Framework versions

Transformers 4.55.2
Pytorch 2.8.0+cu128
Datasets 4.0.0
Tokenizers 0.21.4

craa/exceptions_exp2_swap_0.3_cost_to_hit_1032

exceptions_exp2_swap_0.3_cost_to_hit_1032

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

craa/
exceptions_exp2_swap_0.3_cost_to_hit_1032