fd363df92df929f820cc3ec968ff1e29

This model is a fine-tuned version of albert/albert-base-v1 on the contemmcm/trec dataset. It achieves the following results on the evaluation set:

• Loss: 0.2133
• Data Size: 1.0
• Epoch Runtime: 6.3373
• Accuracy: 0.9604
• F1 Macro: 0.9474

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	1.9228	0	0.7596	0.0854	0.0529
No log	1	170	1.7120	0.0078	1.1270	0.2812	0.0812
No log	2	340	1.6877	0.0156	0.8666	0.4042	0.2455
No log	3	510	1.4664	0.0312	0.9996	0.35	0.2939
No log	4	680	0.9959	0.0625	1.1770	0.7271	0.6073
0.0696	5	850	0.4798	0.125	1.5303	0.8625	0.7252
0.0696	6	1020	0.3803	0.25	2.2449	0.8729	0.8347
0.3216	7	1190	0.2539	0.5	3.7260	0.9375	0.9183
0.2213	8.0	1360	0.2350	1.0	6.4563	0.9437	0.9331
0.1847	9.0	1530	0.1843	1.0	6.4106	0.9563	0.9522
0.137	10.0	1700	0.1985	1.0	6.3932	0.9542	0.9506
0.1168	11.0	1870	0.1987	1.0	6.3907	0.9563	0.9523
0.0712	12.0	2040	0.4379	1.0	6.3647	0.9021	0.8435
0.0562	13.0	2210	0.2133	1.0	6.3373	0.9604	0.9474

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.1