MedRuRobertaLarge_pos

This model is a fine-tuned version of DmitryPogrebnoy/MedRuRobertaLarge on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4867
• Precision: 0.5088
• Recall: 0.5257
• F1: 0.5171
• Accuracy: 0.8997

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	50	0.6663	0.0	0.0	0.0	0.7639
No log	2.0	100	0.5206	0.0178	0.0154	0.0165	0.8015
No log	3.0	150	0.4083	0.0409	0.0617	0.0492	0.8346
No log	4.0	200	0.3900	0.1300	0.2139	0.1617	0.8368
No log	5.0	250	0.3372	0.1893	0.2987	0.2317	0.8598
No log	6.0	300	0.2828	0.2713	0.3622	0.3102	0.8907
No log	7.0	350	0.3583	0.3625	0.4066	0.3833	0.8890
No log	8.0	400	0.2786	0.3638	0.4605	0.4065	0.8995
No log	9.0	450	0.3000	0.3224	0.4181	0.3641	0.8981
0.3576	10.0	500	0.3055	0.4872	0.5145	0.5005	0.9085
0.3576	11.0	550	0.2949	0.4633	0.5106	0.4858	0.9123
0.3576	12.0	600	0.3481	0.4407	0.5723	0.4979	0.9054
0.3576	13.0	650	0.3636	0.4814	0.5241	0.5018	0.9054
0.3576	14.0	700	0.3186	0.4981	0.5010	0.4995	0.9132
0.3576	15.0	750	0.3472	0.4329	0.5780	0.4950	0.9084
0.3576	16.0	800	0.3664	0.4843	0.5665	0.5222	0.9177
0.3576	17.0	850	0.3666	0.4371	0.6089	0.5089	0.9085
0.3576	18.0	900	0.4685	0.4894	0.5356	0.5115	0.9167
0.3576	19.0	950	0.3722	0.4309	0.5703	0.4909	0.9154
0.0824	20.0	1000	0.3861	0.5327	0.5645	0.5482	0.9097
0.0824	21.0	1050	0.6866	0.5201	0.4239	0.4671	0.8853
0.0824	22.0	1100	0.5474	0.4616	0.6493	0.5396	0.8934
0.0824	23.0	1150	0.4203	0.5714	0.5857	0.5785	0.9168
0.0824	24.0	1200	0.4038	0.3748	0.5568	0.4481	0.8989
0.0824	25.0	1250	0.4873	0.5564	0.5414	0.5488	0.9123
0.0824	26.0	1300	0.4516	0.5306	0.5838	0.5560	0.9170
0.0824	27.0	1350	0.4349	0.5738	0.5915	0.5825	0.9110
0.0824	28.0	1400	0.4042	0.5250	0.5857	0.5537	0.9083
0.0824	29.0	1450	0.4187	0.6107	0.6166	0.6136	0.9103
0.0475	30.0	1500	0.3910	0.4615	0.6127	0.5265	0.9060
0.0475	31.0	1550	0.4171	0.5541	0.6416	0.5946	0.9133
0.0475	32.0	1600	0.4948	0.5730	0.6127	0.5922	0.9109
0.0475	33.0	1650	0.4637	0.5048	0.6089	0.5520	0.9118
0.0475	34.0	1700	0.3740	0.5431	0.6185	0.5784	0.9213
0.0475	35.0	1750	0.4047	0.5280	0.5992	0.5614	0.9129
0.0475	36.0	1800	0.4010	0.5352	0.6301	0.5788	0.9150
0.0475	37.0	1850	0.5743	0.5905	0.5530	0.5711	0.9108
0.0475	38.0	1900	0.4936	0.5110	0.4913	0.5010	0.9102
0.0475	39.0	1950	0.4450	0.4537	0.5857	0.5114	0.9119
0.0424	40.0	2000	0.4611	0.4983	0.5588	0.5268	0.9130
0.0424	41.0	2050	0.4748	0.5199	0.5279	0.5239	0.9075
0.0424	42.0	2100	0.5121	0.5264	0.5568	0.5412	0.9126

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.2