Salesforce-codet5-small-CodeXGLUE-CONCODE-test

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

• Loss: 0.8508
• Exact Match: 0.156
• Rouge1: 0.5559
• Rouge2: 0.3857
• Rougel: 0.5378
• Rougelsum: 0.5465
• Bleu: 0.1246

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: 0.001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match	Rouge1	Rouge2	Rougel	Rougelsum	Bleu
1.3563	0.16	500	1.1652	0.1115	0.5098	0.3191	0.4915	0.4982	0.1088
0.9656	0.32	1000	1.0435	0.1245	0.5246	0.3444	0.5075	0.5145	0.1164
0.8627	0.48	1500	0.9851	0.121	0.5275	0.3420	0.5074	0.5154	0.1132
0.7718	0.64	2000	0.9288	0.1385	0.5334	0.3589	0.5174	0.5242	0.1206
0.7237	0.8	2500	0.8867	0.1495	0.5505	0.3762	0.5328	0.5406	0.1208
0.6812	0.96	3000	0.8508	0.156	0.5559	0.3857	0.5378	0.5465	0.1246

Framework versions

• Transformers 4.27.1
• Pytorch 1.12.1+cu113
• Datasets 2.10.1
• Tokenizers 0.13.2