ccb2a5898f7b663d8285e2245ca41c50

This model is a fine-tuned version of Qwen/Qwen2.5-3B on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 4.3285
• Data Size: 1.0
• Epoch Runtime: 71.9306
• Accuracy: 0.75
• F1 Macro: 0.7776
• Rouge1: 0.75
• Rouge2: 0.0
• Rougel: 0.75
• Rougelsum: 0.7507

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	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	8.8729	0	4.8534	0.2578	0.1301	0.2578	0.0	0.2578	0.2578
No log	1	178	8.8059	0.0078	5.2597	0.4645	0.2746	0.4652	0.0	0.4638	0.4645
No log	2	356	4.2649	0.0156	9.2599	0.6513	0.4697	0.6520	0.0	0.6520	0.6520
No log	3	534	4.1950	0.0312	14.2500	0.6527	0.5052	0.6527	0.0	0.6527	0.6520
No log	4	712	3.2819	0.0625	23.1280	0.6896	0.5986	0.6903	0.0	0.6903	0.6896
No log	5	890	3.7243	0.125	31.3457	0.6797	0.5935	0.6797	0.0	0.6804	0.6804
0.2683	6	1068	2.7243	0.25	36.1006	0.7251	0.7132	0.7259	0.0	0.7251	0.7259
2.6769	7	1246	2.6984	0.5	44.1845	0.7656	0.7972	0.7653	0.0	0.7656	0.7656
1.7081	8.0	1424	2.9064	1.0	75.2778	0.7436	0.7553	0.7436	0.0	0.7436	0.7450
1.2819	9.0	1602	3.9140	1.0	69.6241	0.6911	0.6959	0.6914	0.0	0.6911	0.6911
1.009	10.0	1780	4.3888	1.0	67.9576	0.7315	0.7691	0.7315	0.0	0.7315	0.7315
0.6074	11.0	1958	4.3285	1.0	71.9306	0.75	0.7776	0.75	0.0	0.75	0.7507

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1