f7e65a49fd1b64424ee9370116158c3f

This model is a fine-tuned version of deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B on the ccdv/patent-classification [abstract] dataset. It achieves the following results on the evaluation set:

• Loss: 9.8133
• Data Size: 1.0
• Epoch Runtime: 181.2153
• Accuracy: 0.6496
• F1 Macro: 0.6002

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	11.2749	0	10.8773	0.1332	0.0815
No log	1	781	8.1784	0.0078	12.2238	0.2564	0.1565
No log	2	1562	5.6207	0.0156	14.3409	0.4868	0.3293
No log	3	2343	4.8647	0.0312	19.1654	0.5723	0.4255
0.1452	4	3124	4.6690	0.0625	24.6738	0.5897	0.4789
4.5955	5	3905	4.2943	0.125	36.3555	0.6120	0.5563
4.1436	6	4686	4.0531	0.25	58.2511	0.6430	0.5795
3.3867	7	5467	4.0679	0.5	99.3198	0.6573	0.6119
2.3892	8.0	6248	4.3190	1.0	183.3381	0.6605	0.6187
0.5282	9.0	7029	7.4207	1.0	180.1089	0.6693	0.6170
0.5747	10.0	7810	9.8133	1.0	181.2153	0.6496	0.6002

Framework versions

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