Frozen11-50epoch-BERT-multilingual-finetuned-CEFR_ner-10000news

This model is a fine-tuned version of bert-base-multilingual-cased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3983
• Accuracy: 0.2881
• Precision: 0.4062
• Recall: 0.7266
• F1: 0.4014

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.7274	1.0	500	0.6491	0.2455	0.4102	0.4821	0.2722
0.6213	2.0	1000	0.5746	0.2543	0.3790	0.5244	0.2899
0.5375	3.0	1500	0.5073	0.2626	0.3959	0.5531	0.3168
0.4721	4.0	2000	0.4612	0.2690	0.4067	0.5837	0.3466
0.4231	5.0	2500	0.4297	0.2728	0.3975	0.6080	0.3571
0.3828	6.0	3000	0.4104	0.2751	0.4034	0.6224	0.3628
0.3494	7.0	3500	0.3973	0.2769	0.4011	0.6354	0.3658
0.3223	8.0	4000	0.3764	0.2796	0.4004	0.6546	0.3765
0.2991	9.0	4500	0.3693	0.2807	0.4138	0.6578	0.3874
0.2797	10.0	5000	0.3661	0.2811	0.3998	0.6726	0.3811
0.2623	11.0	5500	0.3571	0.2826	0.4118	0.6765	0.3922
0.2476	12.0	6000	0.3514	0.2833	0.4063	0.6877	0.3914
0.2337	13.0	6500	0.3586	0.2828	0.4046	0.6849	0.3880
0.2198	14.0	7000	0.3480	0.2844	0.4107	0.6904	0.3960
0.2096	15.0	7500	0.3495	0.2847	0.4128	0.6893	0.3968
0.2007	16.0	8000	0.3456	0.2852	0.4106	0.7003	0.3984
0.1894	17.0	8500	0.3543	0.2849	0.4003	0.7058	0.3905
0.1816	18.0	9000	0.3532	0.2851	0.4071	0.7066	0.3966
0.1742	19.0	9500	0.3500	0.2857	0.4138	0.7069	0.4024
0.167	20.0	10000	0.3495	0.286	0.4150	0.7079	0.4047
0.159	21.0	10500	0.3599	0.2859	0.4067	0.7093	0.3973
0.1548	22.0	11000	0.3564	0.2863	0.4061	0.7139	0.3980
0.1492	23.0	11500	0.3587	0.2864	0.4081	0.7132	0.3994
0.1433	24.0	12000	0.3607	0.2867	0.4110	0.7148	0.4022
0.1379	25.0	12500	0.3593	0.2871	0.4133	0.7147	0.4045
0.1336	26.0	13000	0.3689	0.2866	0.4062	0.7164	0.3986
0.1296	27.0	13500	0.3656	0.2872	0.4056	0.7207	0.3996
0.1264	28.0	14000	0.3695	0.2871	0.4104	0.7177	0.4029
0.1223	29.0	14500	0.3700	0.2871	0.4113	0.7185	0.4041
0.119	30.0	15000	0.3732	0.2872	0.4086	0.7206	0.4016
0.115	31.0	15500	0.3765	0.2873	0.4096	0.7198	0.4030
0.1126	32.0	16000	0.3738	0.2878	0.4095	0.7239	0.4040
0.11	33.0	16500	0.3825	0.2874	0.4069	0.7224	0.4007
0.1071	34.0	17000	0.3857	0.2874	0.4029	0.7243	0.3976
0.105	35.0	17500	0.3871	0.2874	0.4069	0.7230	0.4008
0.104	36.0	18000	0.3872	0.2875	0.4046	0.7254	0.3997
0.1021	37.0	18500	0.3890	0.2876	0.4063	0.7236	0.4006
0.0997	38.0	19000	0.3886	0.2877	0.4067	0.7259	0.4017
0.0982	39.0	19500	0.3909	0.2877	0.4084	0.7238	0.4027
0.0964	40.0	20000	0.3951	0.2877	0.4076	0.7245	0.4019
0.0948	41.0	20500	0.3945	0.2879	0.4064	0.7258	0.4011
0.0941	42.0	21000	0.3919	0.2881	0.4096	0.7267	0.4044
0.0932	43.0	21500	0.3937	0.2879	0.4066	0.7262	0.4014
0.0922	44.0	22000	0.3965	0.2879	0.4091	0.7261	0.4038
0.0908	45.0	22500	0.3977	0.2880	0.4061	0.7271	0.4013
0.09	46.0	23000	0.3977	0.2880	0.4063	0.7263	0.4014
0.0906	47.0	23500	0.3978	0.2880	0.4051	0.7274	0.4005
0.0893	48.0	24000	0.3981	0.2881	0.4063	0.7269	0.4015
0.0887	49.0	24500	0.3982	0.2881	0.4067	0.7262	0.4017
0.0883	50.0	25000	0.3983	0.2881	0.4062	0.7266	0.4014

Framework versions

• Transformers 4.41.1
• Pytorch 2.2.1
• Datasets 2.19.1
• Tokenizers 0.19.1