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	Accuracy	Validation Loss
4.8415	0.2915	1000	0.2549	4.7474
4.323	0.5831	2000	0.2987	4.2879
4.144	0.8746	3000	0.3147	4.1024
3.996	1.1662	4000	0.3243	3.9912
3.9321	1.4577	5000	0.3312	3.9185
3.8817	1.7493	6000	0.3364	3.8612
3.7381	2.0408	7000	0.3407	3.8162
3.7703	2.3324	8000	0.3436	3.7915
3.7426	2.6239	9000	0.3463	3.7561
3.7392	2.9155	10000	0.3489	3.7314
3.6431	3.2070	11000	0.3506	3.7197
3.6438	3.4985	12000	0.3526	3.7023
3.6547	3.7901	13000	0.3541	3.6818
3.5352	4.0816	14000	0.3555	3.6740
3.5763	4.3732	15000	0.3564	3.6651
3.5867	4.6647	16000	0.3575	3.6524
3.5812	4.9563	17000	0.3588	3.6386
3.5121	5.2478	18000	0.3592	3.6415
3.533	5.5394	19000	0.3602	3.6322
3.538	5.8309	20000	0.3613	3.6200
3.4465	6.1224	21000	0.3618	3.6232
3.4804	6.4140	22000	0.3621	3.6155
3.4915	6.7055	23000	0.3630	3.6050
3.4959	6.9971	24000	0.3638	3.5950
3.4349	7.2886	25000	0.3636	3.6040
3.4475	7.5802	26000	0.3643	3.5957
3.4658	7.8717	27000	0.3652	3.5869
3.3881	8.1633	28000	0.3650	3.5977
3.4187	8.4548	29000	0.3655	3.5894
3.4278	8.7464	30000	0.3665	3.5818
3.3306	9.0379	31000	0.3662	3.5869
3.3772	9.3294	32000	0.3663	3.5856
3.406	9.6210	33000	0.3671	3.5740
3.4154	9.9125	34000	0.3673	3.5680
3.3481	10.2041	35000	0.3673	3.5775
3.3721	10.4956	36000	0.3675	3.5738
3.3873	10.7872	37000	0.3682	3.5640
3.3062	11.0787	38000	0.3679	3.5776
3.3417	11.3703	39000	0.3683	3.5725
3.3709	11.6618	40000	0.3687	3.5621
3.3906	11.9534	41000	0.3693	3.5545
3.3194	12.2449	42000	0.3690	3.5680
3.3376	12.5364	43000	0.3693	3.5594
3.3603	12.8280	44000	0.3697	3.5541
3.2769	13.1195	45000	0.3693	3.5685
3.3208	13.4111	46000	0.3692	3.5613
3.3292	13.7026	47000	0.3703	3.5551
3.3421	13.9942	48000	0.3707	3.5495
3.288	14.2857	49000	0.3699	3.5604
3.3265	14.5773	50000	0.3704	3.5540
3.3345	14.8688	51000	0.3710	3.5460
3.2504	15.1603	52000	0.3701	3.5615
3.2906	15.4519	53000	0.3710	3.5514
3.308	15.7434	54000	0.3713	3.5465
3.219	16.0350	55000	0.3708	3.5575
3.2695	16.3265	56000	0.3710	3.5565
3.2908	16.6181	57000	0.3713	3.5502
3.3106	16.9096	58000	0.3719	3.5402
3.2359	17.2012	59000	0.3710	3.5583
3.262	17.4927	60000	0.3712	3.5551
3.2945	17.7843	61000	0.3719	3.5481
3.2057	18.0758	62000	0.3713	3.5570
3.2388	18.3673	63000	0.3714	3.5548
3.2631	18.6589	64000	0.3720	3.5457
3.2689	18.9504	65000	0.3727	3.5386
3.2134	19.2420	66000	0.3715	3.5542
3.2582	19.5335	67000	0.3718	3.5489
3.2727	19.8251	68000	0.3725	3.5407
3.1987	20.1166	69000	0.3720	3.5575
3.2363	20.4082	70000	0.3721	3.5517
3.2467	20.6997	71000	0.3728	3.5438
3.2545	20.9913	72000	0.3730	3.5363
3.212	21.2828	73000	0.3721	3.5544
3.2347	21.5743	74000	0.3725	3.5484
3.2424	21.8659	75000	0.3733	3.5391
3.1705	22.1574	76000	0.3725	3.5571
3.2067	22.4490	77000	0.3726	3.5532
3.2377	22.7405	78000	0.3731	3.5414
3.1332	23.0321	79000	0.3727	3.5546
3.1869	23.3236	80000	0.3728	3.5519
3.1899	23.6152	81000	3.5575	0.3726
3.2173	23.9067	82000	3.5493	0.3729
3.1754	24.1983	83000	3.5633	0.3724
3.205	24.4898	84000	3.5507	0.3730
3.2061	24.7813	85000	3.5413	0.3736
3.1315	25.0729	86000	3.5547	0.3728
3.1756	25.3644	87000	3.5507	0.3732
3.1993	25.6560	88000	3.5462	0.3736
3.2205	25.9475	89000	3.5371	0.3740
3.1545	26.2391	90000	3.5542	0.3731
3.1863	26.5306	91000	3.5458	0.3737
3.2025	26.8222	92000	3.5389	0.3742
3.1178	27.1137	93000	3.5570	0.3733
3.167	27.4052	94000	3.5496	0.3735
3.1736	27.6968	95000	3.5412	0.3739
3.2045	27.9883	96000	3.5388	0.3742
3.1314	28.2799	97000	3.5566	0.3733
3.1711	28.5714	98000	3.5471	0.3737
3.1802	28.8630	99000	3.5376	0.3745
3.1075	29.1545	100000	3.5586	0.3734

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_3591

exceptions_exp2_swap_0.3_cost_to_hit_3591

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_3591