scenario-kd-pre-ner-full-mdeberta_data-univner_full44

This model is a fine-tuned version of microsoft/mdeberta-v3-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 46.6459
• Precision: 0.8272
• Recall: 0.8335
• F1: 0.8303
• Accuracy: 0.9822

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: 8
• eval_batch_size: 32
• seed: 44
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
149.9231	0.2911	500	110.8748	0.5070	0.1208	0.1951	0.9347
101.1776	0.5822	1000	93.5222	0.7123	0.6586	0.6844	0.9701
89.9222	0.8732	1500	86.5450	0.7487	0.7276	0.7380	0.9748
83.6167	1.1643	2000	81.4135	0.7818	0.7501	0.7656	0.9769
78.8225	1.4554	2500	77.5955	0.7905	0.7547	0.7722	0.9777
75.3094	1.7465	3000	74.1885	0.7825	0.7798	0.7812	0.9783
71.9149	2.0375	3500	71.4168	0.7893	0.8020	0.7956	0.9790
68.8017	2.3286	4000	68.6904	0.8194	0.7778	0.7981	0.9794
66.2935	2.6197	4500	66.3018	0.7981	0.8070	0.8025	0.9802
64.1282	2.9108	5000	64.3227	0.7988	0.8130	0.8059	0.9803
61.983	3.2019	5500	62.6362	0.8141	0.8114	0.8128	0.9808
60.0914	3.4929	6000	60.8145	0.8106	0.8149	0.8127	0.9808
58.497	3.7840	6500	59.2819	0.8126	0.8158	0.8142	0.9812
57.0173	4.0751	7000	58.0187	0.8126	0.7990	0.8058	0.9804
55.5793	4.3662	7500	56.7794	0.8033	0.8240	0.8135	0.9808
54.4031	4.6573	8000	55.5089	0.8072	0.8287	0.8178	0.9812
53.2147	4.9483	8500	54.5450	0.8128	0.8094	0.8111	0.9810
52.0438	5.2394	9000	53.6043	0.8145	0.8222	0.8184	0.9814
51.102	5.5305	9500	52.6326	0.8100	0.8261	0.818	0.9811
50.3841	5.8216	10000	51.8428	0.8138	0.8300	0.8219	0.9815
49.4812	6.1126	10500	51.1615	0.8192	0.8296	0.8244	0.9819
48.7273	6.4037	11000	50.4750	0.8156	0.8201	0.8178	0.9813
48.1157	6.6948	11500	49.8869	0.8190	0.8259	0.8224	0.9818
47.4821	6.9859	12000	49.2946	0.8203	0.8279	0.8241	0.9819
46.889	7.2770	12500	48.8428	0.8178	0.8224	0.8201	0.9816
46.3939	7.5680	13000	48.3821	0.8264	0.8224	0.8244	0.9819
46.0878	7.8591	13500	47.9867	0.8210	0.8272	0.8241	0.9817
45.669	8.1502	14000	47.6715	0.8207	0.8257	0.8232	0.9818
45.3064	8.4413	14500	47.3744	0.8167	0.8336	0.8251	0.9818
45.0768	8.7324	15000	47.1812	0.8221	0.8235	0.8228	0.9821
44.8212	9.0234	15500	46.9769	0.8172	0.8274	0.8223	0.9816
44.6107	9.3145	16000	46.8141	0.8204	0.8298	0.8250	0.9819
44.4495	9.6056	16500	46.7872	0.8189	0.8285	0.8236	0.9819
44.51	9.8967	17000	46.6459	0.8272	0.8335	0.8303	0.9822

Framework versions

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