12ae129b256a4d3323198dc1359fb000

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the ccdv/patent-classification [abstract] dataset. It achieves the following results on the evaluation set:

• Loss: 1.2222
• Data Size: 1.0
• Epoch Runtime: 67.9437
• Accuracy: 0.6408
• F1 Macro: 0.5858

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	2.2315	0	4.4039	0.0312	0.0067
No log	1	781	2.0279	0.0078	5.0424	0.1560	0.0408
No log	2	1562	1.9534	0.0156	5.7586	0.2071	0.0381
No log	3	2343	1.6273	0.0312	7.2381	0.3702	0.2171
0.0419	4	3124	1.4401	0.0625	9.3832	0.4860	0.3120
1.4373	5	3905	1.3822	0.125	13.1942	0.5248	0.3640
1.2733	6	4686	1.2315	0.25	21.2923	0.5661	0.4469
1.1414	7	5467	1.1168	0.5	37.7507	0.6242	0.5339
1.0287	8.0	6248	1.0654	1.0	72.1931	0.6358	0.5602
0.9417	9.0	7029	1.0281	1.0	70.4005	0.6474	0.5744
0.8275	10.0	7810	1.1577	1.0	68.8999	0.6314	0.5679
0.7464	11.0	8591	1.1115	1.0	67.2536	0.6380	0.5802
0.7072	12.0	9372	1.1572	1.0	68.7726	0.6502	0.5917
0.552	13.0	10153	1.2222	1.0	67.9437	0.6408	0.5858

Framework versions

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