e4228f068077b97c74dc4c180cd3fe8a

This model is a fine-tuned version of albert/albert-xxlarge-v2 on the contemmcm/trec dataset. It achieves the following results on the evaluation set:

• Loss: 0.3213
• Data Size: 1.0
• Epoch Runtime: 11.5312
• Accuracy: 0.9583
• F1 Macro: 0.9557

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	2.4008	0	0.8326	0.1479	0.0568
No log	1	170	2.3301	0.0078	1.0394	0.2771	0.0723
No log	2	340	1.7829	0.0156	1.1375	0.2896	0.1016
No log	3	510	1.7716	0.0312	1.6324	0.1771	0.0589
No log	4	680	1.7543	0.0625	1.8707	0.1792	0.0506
0.1018	5	850	1.2161	0.125	2.6145	0.4917	0.3132
0.1018	6	1020	0.5898	0.25	3.8579	0.8417	0.7060
0.3193	7	1190	0.3723	0.5	6.3442	0.9146	0.7694
0.5148	8.0	1360	0.2532	1.0	11.8502	0.9437	0.9423
0.1715	9.0	1530	0.2308	1.0	11.6359	0.9604	0.9638
0.1028	10.0	1700	0.2603	1.0	11.5794	0.9542	0.9548
0.093	11.0	1870	0.2576	1.0	11.5501	0.9625	0.9531
0.0932	12.0	2040	0.2833	1.0	11.5483	0.9583	0.9563
0.0551	13.0	2210	0.3213	1.0	11.5312	0.9583	0.9557

Framework versions

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