lmeninato/t5-small-codesearchnet-multilang-2-archive

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: 0.3690
• Bleu: 0.0229
• Rouge1: 0.4217
• Rouge2: 0.3443
• Avg Length: 17.1574

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	0.4444	0.0215	0.369	0.3113	18.0234
1.954	2.0	750	0.3842	0.0214	0.398	0.3329	16.9644
0.3621	3.0	1125	0.3631	0.0238	0.4089	0.3377	17.3726
0.3166	4.0	1500	0.3472	0.0214	0.4152	0.3413	16.956
0.3166	5.0	1875	0.3403	0.0238	0.4191	0.3437	17.198
0.2846	6.0	2250	0.3333	0.0247	0.414	0.3456	17.5016
0.2605	7.0	2625	0.3357	0.022	0.4215	0.346	17.023
0.2422	8.0	3000	0.3317	0.0287	0.4277	0.3522	17.658
0.2422	9.0	3375	0.3334	0.0252	0.4288	0.351	17.3284
0.2221	10.0	3750	0.3347	0.0225	0.4215	0.3435	17.1506
0.2073	11.0	4125	0.3392	0.0258	0.4234	0.3498	17.4554
0.194	12.0	4500	0.3445	0.0245	0.427	0.3488	17.2398
0.194	13.0	4875	0.3614	0.0253	0.4284	0.3489	17.3902
0.1789	14.0	5250	0.3556	0.0276	0.4261	0.3504	17.5164
0.1681	15.0	5625	0.3690	0.0229	0.4217	0.3443	17.1574

Framework versions

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