05e96e30e07c90287efdcac412d640f4

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

• Loss: 0.2945
• Data Size: 1.0
• Epoch Runtime: 15.7427
• Accuracy: 0.9604
• F1 Macro: 0.9572

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	1.8172	0	1.0057	0.1646	0.0473
No log	1	170	1.7753	0.0078	1.3938	0.2396	0.0961
No log	2	340	1.6908	0.0156	1.5674	0.2917	0.1188
No log	3	510	1.6976	0.0312	2.1838	0.1417	0.0510
No log	4	680	1.5551	0.0625	2.6905	0.4833	0.3492
0.095	5	850	1.0769	0.125	3.7557	0.7292	0.6179
0.095	6	1020	0.4873	0.25	5.9344	0.8688	0.7277
0.4882	7	1190	0.4252	0.5	9.6838	0.9062	0.7600
0.3078	8.0	1360	0.3501	1.0	16.6750	0.9208	0.9013
0.2166	9.0	1530	0.1860	1.0	16.5451	0.9646	0.9684
0.1502	10.0	1700	0.1715	1.0	16.3672	0.9604	0.9524
0.1265	11.0	1870	0.3303	1.0	16.8885	0.9521	0.9583
0.115	12.0	2040	0.3288	1.0	17.1839	0.95	0.9340
0.0564	13.0	2210	0.2000	1.0	16.1723	0.9625	0.9653
0.0914	14.0	2380	0.2945	1.0	15.7427	0.9604	0.9572

Framework versions

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