t5-base-finetuned-c_zh-to-m_zh

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

• Loss: 0.1259
• Bleu: 86.7906
• Gen Len: 7.5595

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	234	0.1445	86.8464	7.5788
No log	2.0	468	0.1357	86.5266	7.6088
0.194	3.0	702	0.1317	86.8389	7.5756
0.194	4.0	936	0.1302	87.0248	7.5648
0.1428	5.0	1170	0.1372	85.5082	7.686
0.1428	6.0	1404	0.1297	86.3936	7.612
0.1328	7.0	1638	0.1273	86.7919	7.5745
0.1328	8.0	1872	0.1266	86.7919	7.5745
0.129	9.0	2106	0.1262	86.9787	7.5606
0.129	10.0	2340	0.1256	86.882	7.5616
0.1262	11.0	2574	0.1259	86.9757	7.5616
0.1262	12.0	2808	0.1255	86.8843	7.5595
0.1262	13.0	3042	0.1257	86.9322	7.5584
0.1262	14.0	3276	0.1256	87.0674	7.5563
0.1238	15.0	3510	0.1259	86.7906	7.5595

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1