yahoo-bert-32shot

This model is a fine-tuned version of bert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.3118
• Accuracy: 0.1289
• Precision: 0.1064
• Recall: 0.1313

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: 100
• eval_batch_size: 100
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 200
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall
1.2358	1.0	1	2.3792	0.09	0.1152	0.0996
1.1843	2.0	2	2.3509	0.1013	0.0961	0.1102
1.155	3.0	3	2.3509	0.1013	0.0961	0.1102
1.1314	4.0	4	2.3370	0.1146	0.0899	0.1221
1.1329	5.0	5	2.3265	0.1179	0.0959	0.1231
1.1008	6.0	6	2.3205	0.1238	0.0951	0.1278
1.0631	7.0	7	2.3157	0.1277	0.0954	0.1297
1.0569	8.0	8	2.3123	0.1306	0.0985	0.1315
1.0326	9.0	9	2.3105	0.129	0.0969	0.1296
1.0291	10.0	10	2.3098	0.1273	0.0951	0.1283
1.0187	11.0	11	2.3094	0.1285	0.1041	0.1295
0.9912	12.0	12	2.3096	0.1296	0.1041	0.1312
0.9704	13.0	13	2.3106	0.1294	0.1084	0.1314
0.9498	14.0	14	2.3118	0.1289	0.1064	0.1313

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1