b85c9ca7555f9f4541a97a534c3089ef

This model is a fine-tuned version of deepseek-ai/DeepSeek-R1-Distill-Qwen-7B on the ccdv/patent-classification [abstract] dataset. It achieves the following results on the evaluation set:

• Loss: 8.4463
• Data Size: 1.0
• Epoch Runtime: 1562.2101
• Accuracy: 0.6304
• F1 Macro: 0.5955

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	31.6335	0	26.4983	0.0264	0.0179
No log	1	781	14.4174	0.0078	37.8868	0.0469	0.0332
No log	2	1562	6.6324	0.0156	55.4921	0.4351	0.2518
No log	3	2343	5.7025	0.0312	87.7690	0.4988	0.3555
0.2297	4	3124	4.8599	0.0625	140.7495	0.5911	0.4622
4.7686	5	3905	4.5438	0.125	239.1684	0.6026	0.5229
4.3478	6	4686	4.1749	0.25	426.5462	0.625	0.5268
3.6198	7	5467	4.1393	0.5	805.8399	0.6520	0.5983
2.692	8.0	6248	4.2240	1.0	1567.2867	0.6601	0.6105
0.9165	9.0	7029	6.5264	1.0	1556.2602	0.6482	0.5901
0.6631	10.0	7810	7.2687	1.0	1550.1707	0.6518	0.6022
0.5347	11.0	8591	8.4463	1.0	1562.2101	0.6304	0.5955

Framework versions

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