1f4ec8e45e310c36305d3f1c935ac1bb

This model is a fine-tuned version of google-bert/bert-large-cased-whole-word-masking-finetuned-squad on the contemmcm/trec dataset. It achieves the following results on the evaluation set:

• Loss: 0.2218
• Data Size: 1.0
• Epoch Runtime: 18.4569
• Accuracy: 0.9667
• F1 Macro: 0.9614

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: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_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: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	2.0720	0	0.9700	0.0187	0.0061
No log	1	170	1.5803	0.0078	1.3161	0.3396	0.1626
No log	2	340	1.4011	0.0156	1.7225	0.4437	0.2593
No log	3	510	0.7209	0.0312	2.4014	0.8417	0.7345
No log	4	680	0.3401	0.0625	3.3093	0.8979	0.8587
0.0417	5	850	0.1367	0.125	4.6615	0.9625	0.9576
0.0417	6	1020	0.2107	0.25	7.0981	0.9542	0.9281
0.2841	7	1190	0.1732	0.5	11.4066	0.9542	0.9520
0.1876	8.0	1360	0.1739	1.0	18.8644	0.9583	0.9626
0.1211	9.0	1530	0.1098	1.0	18.6139	0.975	0.9777
0.089	10.0	1700	0.2985	1.0	19.3343	0.9542	0.9594
0.1156	11.0	1870	0.2728	1.0	18.7886	0.9625	0.9483
0.1054	12.0	2040	0.2144	1.0	18.0334	0.9646	0.9692
0.0457	13.0	2210	0.2218	1.0	18.4569	0.9667	0.9614

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1