Modelpublic

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

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face

0likes13downloads

Model Card

t5-small-codesearchnet-multilang-python

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

—Loss: 0.0682
—Bleu: 0.0401
—Rouge1: 0.6333
—Rouge2: 0.6147
—Avg Length: 16.9514

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
—trainbatchsize: 8
—evalbatchsize: 8
—seed: 42
—gradientaccumulationsteps: 10
—totaltrainbatch_size: 80
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Rouge1	Rouge2	Avg Length
No log	1.0	375	0.0717	0.0407	0.6244	0.6063	17.046
1.6287	2.0	750	0.0589	0.041	0.6321	0.6136	16.9924
0.0592	3.0	1125	0.0551	0.0402	0.6334	0.6152	16.971
0.0511	4.0	1500	0.0542	0.0402	0.6336	0.6155	16.9718
0.0511	5.0	1875	0.0529	0.0401	0.6343	0.6161	16.961
0.0441	6.0	2250	0.0531	0.0402	0.6341	0.6158	16.9626
0.0399	7.0	2625	0.0521	0.0402	0.6337	0.6154	16.97
0.0351	8.0	3000	0.0547	0.0401	0.6341	0.6159	16.964
0.0351	9.0	3375	0.0545	0.0402	0.635	0.6167	16.962
0.0301	10.0	3750	0.0557	0.0402	0.6342	0.6159	16.9646
0.027	11.0	4125	0.0569	0.0402	0.6342	0.6157	16.9622
0.0239	12.0	4500	0.0606	0.0401	0.6342	0.6158	16.9564
0.0239	13.0	4875	0.0616	0.0401	0.6343	0.6163	16.963
0.02	14.0	5250	0.0672	0.0401	0.6336	0.6154	16.9648
0.0185	15.0	5625	0.0682	0.0401	0.6333	0.6147	16.9514

Framework versions

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