CodeT5ForCodeSummary

This model is a fine-tuned version of Salesforce/codet5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8891

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 14400.0
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.0295	1.0	750	0.7786
0.7481	2.0	1500	0.6092
0.6274	3.0	2250	0.5648
0.5562	4.0	3000	0.5395
0.479	5.0	3750	0.5328
0.4297	6.0	4500	0.5227
0.3519	7.0	5250	0.5505
0.3148	8.0	6000	0.5607
0.2535	9.0	6750	0.5909
0.2152	10.0	7500	0.6145
0.1801	11.0	8250	0.6458
0.1635	12.0	9000	0.6921
0.1474	13.0	9750	0.7425
0.1236	14.0	10500	0.7537
0.1075	15.0	11250	0.8059
0.1039	16.0	12000	0.8050
0.0888	17.0	12750	0.8645
0.0857	18.0	13500	0.8714
0.0763	19.0	14250	0.9016
0.048	20.0	15000	0.8891

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.2+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0