scenario-non-kd-scr-ner-full-xlmr_data-univner_half44

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5850
• Precision: 0.3090
• Recall: 0.4141
• F1: 0.3539
• Accuracy: 0.9219

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 44
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.3726	0.5828	500	0.3180	0.3185	0.1437	0.1981	0.9252
0.2842	1.1655	1000	0.3582	0.1651	0.2194	0.1884	0.9023
0.2484	1.7483	1500	0.3015	0.2876	0.1958	0.2330	0.9274
0.2164	2.3310	2000	0.3746	0.1573	0.2725	0.1995	0.8904
0.2037	2.9138	2500	0.3387	0.1816	0.2516	0.2110	0.9029
0.1786	3.4965	3000	0.3330	0.2181	0.2813	0.2457	0.9115
0.1563	4.0793	3500	0.3990	0.1694	0.3402	0.2262	0.8842
0.1285	4.6620	4000	0.3549	0.2149	0.3561	0.2680	0.9025
0.1123	5.2448	4500	0.3587	0.2343	0.3646	0.2852	0.9051
0.0914	5.8275	5000	0.3688	0.2347	0.3747	0.2886	0.9041
0.0742	6.4103	5500	0.3943	0.2424	0.3805	0.2961	0.9068
0.0692	6.9930	6000	0.3877	0.2513	0.3896	0.3055	0.9102
0.0538	7.5758	6500	0.3864	0.2837	0.3865	0.3272	0.9199
0.0463	8.1585	7000	0.4363	0.2537	0.4067	0.3125	0.9082
0.0395	8.7413	7500	0.4260	0.2618	0.4046	0.3179	0.9118
0.0342	9.3240	8000	0.4661	0.2441	0.4050	0.3046	0.9081
0.0309	9.9068	8500	0.4307	0.2850	0.3963	0.3315	0.9195
0.0246	10.4895	9000	0.4781	0.2541	0.4157	0.3154	0.9080
0.0223	11.0723	9500	0.4891	0.2721	0.4035	0.3250	0.9129
0.0182	11.6550	10000	0.4741	0.2897	0.3969	0.3349	0.9190
0.0172	12.2378	10500	0.4542	0.3522	0.3968	0.3732	0.9320
0.0155	12.8205	11000	0.4927	0.2961	0.3885	0.3361	0.9220
0.0127	13.4033	11500	0.4922	0.2991	0.4096	0.3458	0.9196
0.0118	13.9860	12000	0.5092	0.2809	0.4193	0.3364	0.9161
0.0096	14.5688	12500	0.4991	0.3093	0.4059	0.3511	0.9246
0.0095	15.1515	13000	0.5892	0.2645	0.4292	0.3273	0.9095
0.0082	15.7343	13500	0.5076	0.3108	0.4054	0.3519	0.9230
0.007	16.3170	14000	0.5248	0.3094	0.4147	0.3544	0.9228
0.0069	16.8998	14500	0.5352	0.2932	0.4210	0.3457	0.9183
0.006	17.4825	15000	0.5458	0.3034	0.4092	0.3485	0.9205
0.0056	18.0653	15500	0.5786	0.2857	0.4366	0.3454	0.9150
0.004	18.6480	16000	0.5608	0.3048	0.4168	0.3521	0.9214
0.0047	19.2308	16500	0.5394	0.3339	0.4007	0.3643	0.9263
0.0041	19.8135	17000	0.5308	0.3333	0.4085	0.3671	0.9279
0.0037	20.3963	17500	0.5628	0.3081	0.4184	0.3548	0.9207
0.0035	20.9790	18000	0.5735	0.3115	0.4144	0.3556	0.9219
0.0026	21.5618	18500	0.5682	0.3123	0.4109	0.3549	0.9225
0.0027	22.1445	19000	0.5672	0.3264	0.4108	0.3637	0.9255
0.0022	22.7273	19500	0.6179	0.2866	0.4356	0.3457	0.9147
0.0025	23.3100	20000	0.5594	0.3419	0.4004	0.3688	0.9288
0.0021	23.8928	20500	0.5850	0.3090	0.4141	0.3539	0.9219

Framework versions

• Transformers 4.44.2
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.19.1