t5-small-codesearchnet-multilang-python-java-javascript-go

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

• Loss: 0.5955
• Bleu: 0.009
• Rouge1: 0.2321
• Rouge2: 0.0831
• Avg Length: 16.6192

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	0.7349	0.0028	0.1562	0.0364	16.436
2.3117	2.0	750	0.6613	0.0066	0.1818	0.0531	16.824
0.6755	3.0	1125	0.6233	0.007	0.1957	0.0594	16.931
0.5998	4.0	1500	0.6023	0.0082	0.202	0.063	16.7154
0.5998	5.0	1875	0.5925	0.0096	0.2154	0.0703	16.5468
0.5511	6.0	2250	0.5728	0.0091	0.2213	0.0774	15.7216
0.5147	7.0	2625	0.5670	0.0111	0.2311	0.0815	16.6658
0.4861	8.0	3000	0.5628	0.0089	0.2217	0.077	17.038
0.4861	9.0	3375	0.5598	0.0103	0.2311	0.0825	16.362
0.4526	10.0	3750	0.5589	0.0083	0.232	0.086	15.4298
0.4329	11.0	4125	0.5649	0.0098	0.2349	0.0839	16.5468
0.4102	12.0	4500	0.5633	0.0098	0.2366	0.0867	16.4136
0.4102	13.0	4875	0.5841	0.01	0.2385	0.0869	15.9864
0.3841	14.0	5250	0.5777	0.0128	0.2437	0.0894	16.842
0.3673	15.0	5625	0.5955	0.009	0.2321	0.0831	16.6192

Framework versions

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