README.md

metadata

library_name: transformers
license: mit
base_model: microsoft/deberta-v3-base
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - f1
  - precision
  - recall
model-index:
  - name: my-polarization-model
    results: []

my-polarization-model

This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1047
• Accuracy: 0.6357
• F1: 0.4941
• Precision: 0.4041
• Recall: 0.6357

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-07
• train_batch_size: 100
• eval_batch_size: 100
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 200

Training results

Training Loss	Epoch	Step	Accuracy	F1	Validation Loss	Precision	Recall
1.2002	3.8462	100	0.3643	0.1946	1.1760	0.1327	0.3643
1.1958	7.6923	200	0.3643	0.1946	1.1699	0.1327	0.3643
1.1865	11.5385	300	0.3643	0.1946	1.1643	0.1327	0.3643
1.1827	15.3846	400	0.3643	0.1946	1.1590	0.1327	0.3643
1.1763	19.2308	500	0.3643	0.1946	1.1541	0.1327	0.3643
1.1697	23.0769	600	0.3814	0.2352	1.1496	0.6857	0.3814
1.1686	26.9231	700	0.5566	0.5531	1.1454	0.6590	0.5566
1.1653	30.7692	800	0.6496	0.5773	1.1415	0.6270	0.6496
1.1614	34.6154	900	0.6372	0.5031	1.1379	0.6238	0.6372
1.1589	38.4615	1000	0.6341	0.4933	1.1347	0.4037	0.6341
1.1524	42.3077	1100	0.6357	0.4941	1.1316	0.4041	0.6357
1.1472	46.1538	1200	0.6357	0.4941	1.1288	0.4041	0.6357
1.1465	50.0	1300	0.6357	0.4941	1.1263	0.4041	0.6357
1.1479	53.8462	1400	0.6357	0.4941	1.1240	0.4041	0.6357
1.147	57.6923	1500	0.6357	0.4941	1.1219	0.4041	0.6357
1.1489	61.5385	1600	0.6357	0.4941	1.1201	0.4041	0.6357
1.1421	65.3846	1700	0.6357	0.4941	1.1184	0.4041	0.6357
1.1424	69.2308	1800	0.6357	0.4941	1.1169	0.4041	0.6357
1.1363	73.0769	1900	0.6357	0.4941	1.1156	0.4041	0.6357
1.14	76.9231	2000	0.6357	0.4941	1.1144	0.4041	0.6357
1.1399	80.7692	2100	0.6357	0.4941	1.1134	0.4041	0.6357
1.1403	84.6154	2200	0.6357	0.4941	1.1124	0.4041	0.6357
1.1417	88.4615	2300	0.6357	0.4941	1.1115	0.4041	0.6357
1.1352	92.3077	2400	0.6357	0.4941	1.1108	0.4041	0.6357
1.127	96.1538	2500	0.6357	0.4941	1.1101	0.4041	0.6357
1.1245	100.0	2600	0.6357	0.4941	1.1095	0.4041	0.6357
1.1309	103.8462	2700	0.6357	0.4941	1.1090	0.4041	0.6357
1.1318	107.6923	2800	0.6357	0.4941	1.1085	0.4041	0.6357
1.1293	111.5385	2900	0.6357	0.4941	1.1080	0.4041	0.6357
1.1315	115.3846	3000	0.6357	0.4941	1.1076	0.4041	0.6357
1.1299	119.2308	3100	0.6357	0.4941	1.1073	0.4041	0.6357
1.1314	123.0769	3200	0.6357	0.4941	1.1070	0.4041	0.6357
1.1309	126.9231	3300	0.6357	0.4941	1.1067	0.4041	0.6357
1.1235	130.7692	3400	0.6357	0.4941	1.1064	0.4041	0.6357
1.1367	134.6154	3500	0.6357	0.4941	1.1062	0.4041	0.6357
1.1362	138.4615	3600	0.6357	0.4941	1.1060	0.4041	0.6357
1.1194	142.3077	3700	0.6357	0.4941	1.1058	0.4041	0.6357
1.1283	146.1538	3800	0.6357	0.4941	1.1057	0.4041	0.6357
1.1183	150.0	3900	0.6357	0.4941	1.1055	0.4041	0.6357
1.1252	153.8462	4000	0.6357	0.4941	1.1054	0.4041	0.6357
1.1357	157.6923	4100	0.6357	0.4941	1.1053	0.4041	0.6357
1.132	161.5385	4200	0.6357	0.4941	1.1052	0.4041	0.6357
1.1292	165.3846	4300	0.6357	0.4941	1.1051	0.4041	0.6357
1.1302	169.2308	4400	0.6357	0.4941	1.1050	0.4041	0.6357
1.1282	173.0769	4500	0.6357	0.4941	1.1049	0.4041	0.6357
1.1323	176.9231	4600	1.1049	0.6357	0.4941	0.4041	0.6357
1.1368	180.7692	4700	1.1048	0.6357	0.4941	0.4041	0.6357
1.1364	184.6154	4800	1.1048	0.6357	0.4941	0.4041	0.6357
1.1207	188.4615	4900	1.1048	0.6357	0.4941	0.4041	0.6357
1.1302	192.3077	5000	1.1048	0.6357	0.4941	0.4041	0.6357
1.1235	196.1538	5100	1.1047	0.6357	0.4941	0.4041	0.6357
1.1236	200.0	5200	1.1047	0.6357	0.4941	0.4041	0.6357

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.4.1
• Tokenizers 0.22.1