bert_classifier_sped_transactions

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: 0.1678
• Precision: 0.6500
• Recall: 0.3266
• F1 Unweighted: 0.4347
• F1 Weighted: 0.9451
• F05 Unweighted: 0.5426
• F05 Weighted: 0.9432
• Pr Auc: 0.4367
• Roc Auc: 0.6581

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: 8
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use 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: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1 Unweighted	F1 Weighted	F05 Unweighted	F05 Weighted	Pr Auc	Roc Auc
0.1772	0.0864	5000	0.1694	0.6414	0.3224	0.4291	0.9445	0.5354	0.9425	0.4278	0.6559
0.2169	0.1728	10000	0.2158	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2124	0.2593	15000	0.2157	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0560	0.5
0.2211	0.3457	20000	0.2176	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2149	0.4321	25000	0.2155	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2146	0.5185	30000	0.2177	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2167	0.6050	35000	0.2176	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2143	0.6914	40000	0.2166	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2187	0.7778	45000	0.2156	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2167	0.8642	50000	0.2157	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.22	0.9507	55000	0.2160	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2146	1.0371	60000	0.2158	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2142	1.1235	65000	0.2172	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2205	1.2100	70000	0.2153	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2244	1.2964	75000	0.2154	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0566	0.5
0.2199	1.3828	80000	0.2153	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.215	1.4692	85000	0.2188	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2091	1.5557	90000	0.2169	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2136	1.6421	95000	0.2189	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0557	0.5
0.2127	1.7285	100000	0.2210	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2191	1.8149	105000	0.2157	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0561	0.5
0.2189	1.9014	110000	0.2157	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2189	1.9878	115000	0.2155	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2123	2.0742	120000	0.2162	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0555	0.5
0.2114	2.1606	125000	0.2158	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0564	0.5
0.2166	2.2471	130000	0.2153	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2209	2.3335	135000	0.2154	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2047	2.4199	140000	0.2169	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0559	0.5
0.2173	2.5063	145000	0.2162	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2235	2.5928	150000	0.2154	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2069	2.6792	155000	0.2163	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2166	2.7656	160000	0.2158	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2165	2.8520	165000	0.2157	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5
0.2113	2.9385	170000	0.2160	0.0	0.0	0.0	0.9171	0.0	0.9015	0.0558	0.5

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1
• Datasets 3.1.0
• Tokenizers 0.20.3