dd8d53b7973c977c950d64eddf46d366

This model is a fine-tuned version of facebook/opt-350m on the ccdv/patent-classification [abstract] dataset. It achieves the following results on the evaluation set:

• Loss: 1.4562
• Data Size: 1.0
• Epoch Runtime: 115.4938
• Accuracy: 0.6128
• F1 Macro: 0.5753

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	3.3611	0	7.9595	0.1060	0.0721
No log	1	781	2.2497	0.0078	8.9394	0.2222	0.0416
No log	2	1562	1.6259	0.0156	9.8390	0.3782	0.2144
No log	3	2343	1.4692	0.0312	12.1509	0.4597	0.3075
0.0406	4	3124	1.2921	0.0625	15.9530	0.5623	0.4478
1.2451	5	3905	1.2038	0.125	22.9115	0.5675	0.4956
1.169	6	4686	1.0811	0.25	36.2092	0.6218	0.5401
1.0054	7	5467	1.1172	0.5	63.9539	0.6310	0.5552
0.896	8.0	6248	1.0867	1.0	115.7733	0.6364	0.5812
0.7385	9.0	7029	1.2085	1.0	115.4920	0.6366	0.5885
0.465	10.0	7810	1.4562	1.0	115.4938	0.6128	0.5753

Framework versions

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