t5-small-codesearchnet-multilang-python-java

This model is a fine-tuned version of t5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7015
• Bleu: 0.0045
• Rouge1: 0.2194
• Rouge2: 0.0741
• Avg Length: 15.9976

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
• gradient_accumulation_steps: 10
• total_train_batch_size: 80
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Rouge1	Rouge2	Avg Length
No log	1.0	375	0.9005	0.0013	0.1397	0.0334	16.3976
2.3568	2.0	750	0.8036	0.0023	0.1737	0.0526	15.8896
0.7576	3.0	1125	0.7584	0.0021	0.1856	0.0558	15.3102
0.6778	4.0	1500	0.7298	0.0024	0.1922	0.0597	15.3544
0.6778	5.0	1875	0.7114	0.0037	0.2114	0.0704	15.7588
0.6206	6.0	2250	0.6949	0.0039	0.2093	0.0729	15.8088
0.5856	7.0	2625	0.6927	0.0042	0.2143	0.0711	16.5838
0.5447	8.0	3000	0.6867	0.005	0.2151	0.0717	17.2174
0.5447	9.0	3375	0.6895	0.0043	0.2179	0.0736	16.1068
0.5117	10.0	3750	0.6876	0.0038	0.2229	0.0777	15.5094
0.4892	11.0	4125	0.6800	0.0047	0.2201	0.0783	16.6902
0.4629	12.0	4500	0.6903	0.0047	0.2203	0.0771	16.7658
0.4629	13.0	4875	0.6947	0.0056	0.227	0.0777	16.8108
0.4355	14.0	5250	0.6999	0.0027	0.2028	0.0715	15.6776
0.418	15.0	5625	0.7015	0.0045	0.2194	0.0741	15.9976

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3