davidgaofc/SFT_shadow

shadow

This model is a fine-tuned version of t5-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2249
• Rouge1: 0.2638
• Rouge2: 0.1186
• Rougel: 0.2171
• Rougelsum: 0.2169
• Gen Len: 146.8555

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.6504	1.0	718	1.5109	0.0112	0.0026	0.0098	0.0098	13.3183
1.7193	2.0	1436	1.3836	0.1697	0.0487	0.1355	0.1354	140.8933
1.5492	3.0	2154	1.3174	0.202	0.0564	0.1591	0.1592	153.2811
1.51	4.0	2872	1.2829	0.2152	0.0691	0.1708	0.1709	156.3921
1.4341	5.0	3590	1.2613	0.2336	0.0895	0.1866	0.1866	156.2713
1.4335	6.0	4308	1.2466	0.2534	0.1118	0.2073	0.2075	146.7207
1.4033	7.0	5026	1.2363	0.2591	0.1168	0.213	0.213	144.5354
1.4045	8.0	5744	1.2298	0.261	0.1171	0.2149	0.2149	147.7884
1.3839	9.0	6462	1.2260	0.263	0.1178	0.216	0.2159	147.3518
1.3863	10.0	7180	1.2249	0.2638	0.1186	0.2171	0.2169	146.8555

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.1+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0