a04f436e4d4e5de5b76ba7fd904c3c78

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

• Loss: 1.3133
• Data Size: 1.0
• Epoch Runtime: 116.4855
• Accuracy: 0.6518
• F1 Macro: 0.5961

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	2.1164	0	7.2057	0.1510	0.0292
No log	1	781	1.9938	0.0078	8.4663	0.2718	0.0863
No log	2	1562	1.5860	0.0156	9.4765	0.3872	0.1908
No log	3	2343	1.4555	0.0312	11.5807	0.4683	0.3026
0.0374	4	3124	1.3472	0.0625	15.1891	0.5483	0.4418
1.3362	5	3905	1.1729	0.125	21.9533	0.5645	0.5232
1.2058	6	4686	1.0861	0.25	36.7036	0.6178	0.5091
1.0744	7	5467	1.1440	0.5	63.8407	0.6174	0.5373
0.9986	8.0	6248	1.1125	1.0	121.1422	0.6314	0.5654
0.9276	9.0	7029	1.0343	1.0	123.2721	0.6502	0.5854
0.8526	10.0	7810	1.1100	1.0	119.2637	0.6482	0.5935
0.7286	11.0	8591	1.0734	1.0	118.6671	0.6601	0.6068
0.674	12.0	9372	1.2207	1.0	117.9967	0.6583	0.6129
0.558	13.0	10153	1.3133	1.0	116.4855	0.6518	0.5961

Framework versions

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