trocr-small-printedkorean-deleteunusedchar_denoise

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.8528
• Cer: 0.3241
• Wer: 0.3436

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: 4e-05
• train_batch_size: 64
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.9789	0.11	1000	1.5260	0.2830	0.2979
1.0254	0.21	2000	1.5167	0.2827	0.2991
1.0925	0.32	3000	1.5093	0.2879	0.3036
1.0925	0.43	4000	1.4969	0.2872	0.3037
1.0883	0.53	5000	1.4893	0.2848	0.2998
1.0411	0.64	6000	1.4839	0.2832	0.2994
1.1372	0.75	7000	1.4771	0.2841	0.2993
1.1086	0.85	8000	1.4772	0.2838	0.3010
1.1658	0.96	9000	1.4725	0.2843	0.3013
0.9649	1.07	10000	1.5067	0.2825	0.2978
0.891	1.17	11000	1.5058	0.2846	0.2993
1.0862	1.28	12000	1.4938	0.2873	0.3001
1.0845	1.39	13000	1.4923	0.2867	0.3030
1.0306	1.49	14000	1.4905	0.2833	0.2976
0.9651	1.6	15000	1.4948	0.2835	0.2986
1.1189	1.71	16000	1.4811	0.2856	0.3018
1.0727	1.82	17000	1.4763	0.2864	0.3010
1.0587	1.92	18000	1.4782	0.2830	0.2994
0.9599	2.03	19000	1.5071	0.2866	0.3011
1.0625	2.14	20000	1.5029	0.2837	0.2991
1.1031	2.24	21000	1.5084	0.2880	0.3037
0.9586	2.35	22000	1.5068	0.2842	0.2967
1.0012	2.46	23000	1.4990	0.2861	0.3017
1.0871	2.56	24000	1.4905	0.2848	0.2989
0.9501	2.67	25000	1.4848	0.2835	0.2985
0.9761	2.78	26000	1.4860	0.2819	0.2983
1.0241	2.88	27000	1.4728	0.2813	0.2949
1.065	2.99	28000	1.4826	0.2829	0.2976
1.0412	3.1	29000	1.5160	0.2879	0.3022
0.9972	3.2	30000	1.5105	0.2867	0.3009
0.9758	3.31	31000	1.5037	0.2827	0.2962
0.9521	3.42	32000	1.4986	0.2824	0.2975
1.0124	3.52	33000	1.4959	0.2870	0.3022
0.9437	3.63	34000	1.4936	0.2837	0.2985
0.9977	3.74	35000	1.4901	0.2827	0.2976
1.045	3.84	36000	1.4896	0.2836	0.2995
1.0193	3.95	37000	1.4796	0.2853	0.2968
0.9276	4.06	38000	1.5135	0.2836	0.2974
0.9583	4.16	39000	1.5174	0.2825	0.2956
1.0952	4.27	40000	1.5143	0.2889	0.3026
1.0029	4.38	41000	1.5124	0.2835	0.2986
1.0129	4.48	42000	1.5062	0.2858	0.3006
1.012	4.59	43000	1.5096	0.2845	0.2997
0.9478	4.7	44000	1.5008	0.2853	0.3002
1.0272	4.8	45000	1.5012	0.2825	0.2958
1.0532	4.91	46000	1.4957	0.2832	0.2974
0.921	5.02	47000	1.5206	0.2871	0.3009
0.9246	5.12	48000	1.5259	0.2832	0.2983
0.9689	5.23	49000	1.5274	0.2842	0.2985
1.127	5.34	50000	1.5186	0.2874	0.2997
0.9616	5.45	51000	1.5200	0.2839	0.2971
0.9833	5.55	52000	1.5142	0.2853	0.2980
0.9113	5.66	53000	1.5172	0.2828	0.2972
0.9753	5.77	54000	1.5032	0.2829	0.2944
0.9159	5.87	55000	1.4970	0.2850	0.2987
1.0663	5.98	56000	1.4979	0.2850	0.2985
0.966	6.09	57000	1.5263	0.2862	0.2976
0.9234	6.19	58000	1.5342	0.2824	0.2955
0.9234	6.3	59000	1.5327	0.2838	0.2968
0.972	6.41	60000	1.5291	0.2849	0.2983
1.0319	6.51	61000	1.5329	0.2836	0.2963
0.9746	6.62	62000	1.5222	0.2838	0.2972
1.0719	6.73	63000	1.5249	0.2848	0.2967
1.0695	6.83	64000	1.5183	0.2882	0.3013
1.0271	6.94	65000	1.5119	0.2859	0.2982
1.0119	7.05	66000	1.5410	0.2881	0.3014
0.9341	7.15	67000	1.5489	0.2871	0.2995
0.9544	7.26	68000	1.5443	0.2836	0.2974
1.0198	7.37	69000	1.5401	0.2844	0.2983
0.9269	7.47	70000	1.5384	0.2823	0.2948
0.9767	7.58	71000	1.5375	0.2853	0.2987
0.9861	7.69	72000	1.5282	0.2872	0.2991
1.0539	7.79	73000	1.5244	0.2854	0.2974
0.8879	7.9	74000	1.5305	0.2851	0.2989
0.8739	8.01	75000	1.5454	0.2868	0.2998
1.0469	8.11	76000	1.5533	0.2908	0.3033
0.9449	8.22	77000	1.5586	0.2886	0.3007
0.987	8.33	78000	1.5478	0.2900	0.3029
0.9088	8.43	79000	1.5541	0.2848	0.2976
0.8866	8.54	80000	1.5456	0.2868	0.2986
1.0161	8.65	81000	1.5494	0.2889	0.3002
0.982	8.76	82000	1.5388	0.2842	0.2976
0.956	8.86	83000	1.5377	0.2875	0.3010
0.9267	8.97	84000	1.5415	0.2870	0.2982
0.8889	9.08	85000	1.5729	0.2855	0.2983
0.9518	9.18	86000	1.5719	0.2899	0.3030
0.9469	9.29	87000	1.5661	0.2887	0.3020
0.9614	9.4	88000	1.5661	0.2938	0.3075
0.9073	9.5	89000	1.5620	0.2889	0.3021
0.9808	9.61	90000	1.5706	0.2884	0.3006
0.93	9.72	91000	1.5614	0.2923	0.3056
0.969	9.82	92000	1.5585	0.2891	0.3022
0.912	9.93	93000	1.5562	0.2890	0.3002
0.9301	10.04	94000	1.5802	0.2910	0.3052
0.9265	10.14	95000	1.5834	0.2902	0.3030
0.874	10.25	96000	1.5792	0.2898	0.3025
0.877	10.36	97000	1.5813	0.2892	0.3009
0.9479	10.46	98000	1.5716	0.2859	0.2991
1.1058	10.57	99000	1.5676	0.2964	0.3088
0.8595	10.68	100000	1.5734	0.2865	0.2999
0.924	10.78	101000	1.5721	0.2911	0.3024
0.9618	10.89	102000	1.5665	0.2885	0.3017
0.958	11.0	103000	1.5662	0.2870	0.2998
0.8141	11.1	104000	1.5915	0.2927	0.3041
0.9141	11.21	105000	1.5912	0.2952	0.3076
0.8718	11.32	106000	1.5941	0.2897	0.3034
0.9337	11.42	107000	1.5873	0.2927	0.3041
0.9101	11.53	108000	1.5901	0.2931	0.3064
0.9166	11.64	109000	1.5813	0.2885	0.3021
0.9369	11.74	110000	1.5768	0.2924	0.3037
1.0187	11.85	111000	1.5823	0.2884	0.2998
0.8388	11.96	112000	1.5763	0.2915	0.3041
0.9172	12.06	113000	1.6043	0.2923	0.3055
0.8312	12.17	114000	1.6103	0.2950	0.3100
0.8414	12.28	115000	1.6086	0.2913	0.3030
0.8941	12.39	116000	1.6057	0.2960	0.3072
0.9055	12.49	117000	1.6044	0.2905	0.3032
0.9492	12.6	118000	1.6020	0.2905	0.3030
0.9831	12.71	119000	1.6003	0.2906	0.3030
0.779	12.81	120000	1.6006	0.2892	0.3029
0.9086	12.92	121000	1.5959	0.2914	0.3038
0.866	13.03	122000	1.6154	0.2893	0.3045
0.7936	13.13	123000	1.6200	0.2951	0.3088
0.8693	13.24	124000	1.6229	0.2934	0.3075
0.8463	13.35	125000	1.6211	0.2957	0.3092
0.8946	13.45	126000	1.6192	0.2957	0.3095
0.8616	13.56	127000	1.6124	0.2928	0.3045
0.88	13.67	128000	1.6147	0.2940	0.3068
0.938	13.77	129000	1.6067	0.2904	0.3055
0.8949	13.88	130000	1.6065	0.2922	0.3061
0.8911	13.99	131000	1.5997	0.2896	0.3034
0.813	14.09	132000	1.6342	0.2915	0.3042
0.8077	14.2	133000	1.6314	0.2926	0.3053
0.8068	14.31	134000	1.6316	0.2973	0.3106
0.828	14.41	135000	1.6379	0.2899	0.3032
0.9271	14.52	136000	1.6351	0.2958	0.3095
0.8314	14.63	137000	1.6294	0.2984	0.3127
0.8468	14.73	138000	1.6318	0.2951	0.3091
0.8859	14.84	139000	1.6307	0.2944	0.3068
0.9133	14.95	140000	1.6172	0.2945	0.3071
0.8673	15.05	141000	1.6481	0.3022	0.3164
0.7975	15.16	142000	1.6517	0.2966	0.3091
0.7824	15.27	143000	1.6588	0.2982	0.3107
0.741	15.37	144000	1.6534	0.2936	0.3079
0.8285	15.48	145000	1.6464	0.2970	0.3112
0.8817	15.59	146000	1.6431	0.2955	0.3096
0.807	15.7	147000	1.6395	0.3005	0.3141
0.8341	15.8	148000	1.6407	0.2956	0.3087
0.8734	15.91	149000	1.6349	0.3031	0.3158
0.7333	16.02	150000	1.6556	0.2980	0.3123
0.7652	16.12	151000	1.6663	0.3046	0.3191
0.8292	16.23	152000	1.6680	0.3017	0.3165
0.9419	16.34	153000	1.6651	0.3053	0.3195
0.8787	16.44	154000	1.6635	0.3008	0.3156
0.9263	16.55	155000	1.6578	0.2975	0.3100
0.8244	16.66	156000	1.6601	0.2990	0.3135
0.8557	16.76	157000	1.6579	0.2949	0.3081
0.8749	16.87	158000	1.6500	0.2986	0.3121
0.7865	16.98	159000	1.6466	0.3024	0.3166
0.7753	17.08	160000	1.6764	0.3046	0.3185
0.8026	17.19	161000	1.6789	0.2995	0.3133
0.8223	17.3	162000	1.6821	0.3026	0.3177
0.7264	17.4	163000	1.6750	0.3035	0.3166
0.8394	17.51	164000	1.6716	0.3001	0.3139
0.7711	17.62	165000	1.6698	0.3027	0.3162
0.8279	17.72	166000	1.6672	0.2968	0.3133
0.8186	17.83	167000	1.6714	0.2980	0.3123
0.819	17.94	168000	1.6609	0.2991	0.3139
0.7633	18.04	169000	1.6941	0.3046	0.3181
0.8339	18.15	170000	1.6916	0.3154	0.3305
0.7452	18.26	171000	1.6963	0.3085	0.3235
0.8282	18.36	172000	1.7003	0.3079	0.3220
0.789	18.47	173000	1.6979	0.3037	0.3187
0.8215	18.58	174000	1.6940	0.3154	0.3297
0.7891	18.68	175000	1.6897	0.3040	0.3193
0.8306	18.79	176000	1.6898	0.3045	0.3195
0.8575	18.9	177000	1.6850	0.2985	0.3142
0.784	19.0	178000	1.7042	0.3013	0.3170
0.7566	19.11	179000	1.7094	0.3057	0.3223
0.7863	19.22	180000	1.7119	0.3039	0.3223
0.8157	19.33	181000	1.7098	0.3116	0.3266
0.8782	19.43	182000	1.7091	0.3061	0.3232
0.7863	19.54	183000	1.7071	0.3033	0.3185
0.8008	19.65	184000	1.7084	0.3077	0.3254
0.7919	19.75	185000	1.6996	0.3028	0.3204
0.8435	19.86	186000	1.7012	0.3058	0.3204
0.87	19.97	187000	1.7021	0.3046	0.3201
0.6902	20.07	188000	1.7367	0.2996	0.3160
0.7449	20.18	189000	1.7309	0.3119	0.3293
0.8253	20.29	190000	1.7297	0.3095	0.3254
0.7288	20.39	191000	1.7273	0.3050	0.3220
0.776	20.5	192000	1.7294	0.3049	0.3228
0.7632	20.61	193000	1.7238	0.3179	0.3349
0.7496	20.71	194000	1.7257	0.3107	0.3253
0.8491	20.82	195000	1.7209	0.3088	0.3253
0.7668	20.93	196000	1.7177	0.3080	0.3238
0.6755	21.03	197000	1.7415	0.3087	0.3258
0.6846	21.14	198000	1.7406	0.3059	0.3218
0.7551	21.25	199000	1.7416	0.3167	0.3319
0.7083	21.35	200000	1.7429	0.3120	0.3267
0.7631	21.46	201000	1.7434	0.3074	0.3240
0.7814	21.57	202000	1.7408	0.3069	0.3247
0.7646	21.67	203000	1.7395	0.3109	0.3292
0.7329	21.78	204000	1.7373	0.3149	0.3331
0.7393	21.89	205000	1.7377	0.3077	0.3242
0.766	21.99	206000	1.7351	0.3097	0.3259
0.696	22.1	207000	1.7593	0.3116	0.3266
0.8049	22.21	208000	1.7660	0.3153	0.3340
0.8011	22.31	209000	1.7611	0.3164	0.3331
0.7884	22.42	210000	1.7613	0.3127	0.3298
0.7607	22.53	211000	1.7633	0.3120	0.3300
0.7159	22.64	212000	1.7603	0.3080	0.3258
0.6687	22.74	213000	1.7571	0.3120	0.3297
0.7331	22.85	214000	1.7529	0.3060	0.3227
0.8109	22.96	215000	1.7582	0.3157	0.3331
0.7407	23.06	216000	1.7757	0.3191	0.3347
0.69	23.17	217000	1.7770	0.3162	0.3343
0.6752	23.28	218000	1.7762	0.3255	0.3437
0.7192	23.38	219000	1.7751	0.3161	0.3337
0.742	23.49	220000	1.7782	0.3136	0.3319
0.7184	23.6	221000	1.7771	0.3225	0.3398
0.7296	23.7	222000	1.7796	0.3287	0.3452
0.7527	23.81	223000	1.7722	0.3172	0.3335
0.7093	23.92	224000	1.7713	0.3128	0.3306
0.6691	24.02	225000	1.7913	0.3197	0.3362
0.6767	24.13	226000	1.7907	0.3153	0.3339
0.7133	24.24	227000	1.7925	0.3252	0.3436
0.7094	24.34	228000	1.7910	0.3197	0.3372
0.7848	24.45	229000	1.7918	0.3263	0.3437
0.7146	24.56	230000	1.7903	0.3202	0.3375
0.6932	24.66	231000	1.7895	0.3183	0.3345
0.6632	24.77	232000	1.7895	0.3193	0.3355
0.7587	24.88	233000	1.7894	0.3215	0.3387
0.7197	24.98	234000	1.7863	0.3192	0.3378
0.7759	25.09	235000	1.8054	0.3178	0.3371
0.6729	25.2	236000	1.8069	0.3165	0.3335
0.6144	25.3	237000	1.8059	0.3143	0.3317
0.6838	25.41	238000	1.8077	0.3183	0.3362
0.6925	25.52	239000	1.8017	0.3271	0.3432
0.7301	25.62	240000	1.8035	0.3206	0.3375
0.7169	25.73	241000	1.8043	0.3198	0.3370
0.6717	25.84	242000	1.8042	0.3215	0.3364
0.7113	25.94	243000	1.8048	0.3164	0.3329
0.6544	26.05	244000	1.8186	0.3225	0.3375
0.6664	26.16	245000	1.8216	0.3237	0.3403
0.7131	26.27	246000	1.8234	0.3201	0.3389
0.6202	26.37	247000	1.8222	0.3193	0.3355
0.7311	26.48	248000	1.8173	0.3221	0.3383
0.6377	26.59	249000	1.8200	0.3178	0.3356
0.6931	26.69	250000	1.8190	0.3251	0.3409
0.6402	26.8	251000	1.8198	0.3209	0.3385
0.6492	26.91	252000	1.8188	0.3209	0.3390
0.6544	27.01	253000	1.8279	0.3212	0.3380
0.641	27.12	254000	1.8321	0.3221	0.3405
0.6474	27.23	255000	1.8335	0.3230	0.3407
0.6556	27.33	256000	1.8346	0.3216	0.3395
0.6917	27.44	257000	1.8332	0.3250	0.3422
0.6741	27.55	258000	1.8344	0.3228	0.3399
0.6758	27.65	259000	1.8338	0.3182	0.3362
0.6579	27.76	260000	1.8353	0.3194	0.3383
0.6836	27.87	261000	1.8334	0.3235	0.3405
0.6936	27.97	262000	1.8335	0.3192	0.3391
0.6588	28.08	263000	1.8444	0.3197	0.3386
0.6477	28.19	264000	1.8432	0.3189	0.3403
0.6531	28.29	265000	1.8460	0.3243	0.3422
0.6467	28.4	266000	1.8448	0.3208	0.3390
0.6592	28.51	267000	1.8446	0.3234	0.3391
0.5774	28.61	268000	1.8462	0.3261	0.3446
0.6444	28.72	269000	1.8468	0.3222	0.3401
0.6853	28.83	270000	1.8459	0.3224	0.3398
0.6406	28.93	271000	1.8448	0.3185	0.3362
0.705	29.04	272000	1.8535	0.3243	0.3426
0.6171	29.15	273000	1.8535	0.3210	0.3398
0.651	29.25	274000	1.8536	0.3263	0.3459
0.6771	29.36	275000	1.8533	0.3238	0.3426
0.6554	29.47	276000	1.8536	0.3225	0.3421
0.6585	29.58	277000	1.8523	0.3212	0.3410
0.5784	29.68	278000	1.8531	0.3187	0.3379
0.6542	29.79	279000	1.8530	0.3242	0.3430
0.6015	29.9	280000	1.8528	0.3241	0.3436

Framework versions

• Transformers 4.28.0
• Pytorch 1.13.1+cu116
• Datasets 2.14.4
• Tokenizers 0.13.3