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: 0.8169
• Bleu: 0.0012
• Rouge1: 0.1986
• Rouge2: 0.0594
• Avg Length: 14.004

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	0.9943	0.0003	0.1637	0.0365	13.785
2.445	2.0	750	0.8991	0.0002	0.171	0.041	13.0266
0.8324	3.0	1125	0.8509	0.001	0.1931	0.0499	14.9474
0.7567	4.0	1500	0.8184	0.0015	0.2019	0.0561	14.9598
0.7567	5.0	1875	0.8002	0.0016	0.2097	0.0608	14.496
0.6947	6.0	2250	0.7793	0.0016	0.2138	0.0631	14.6502
0.658	7.0	2625	0.7721	0.0018	0.2104	0.0617	15.2
0.6186	8.0	3000	0.7669	0.0023	0.2175	0.0642	15.7472
0.6186	9.0	3375	0.7792	0.0027	0.2218	0.0664	15.862
0.58	10.0	3750	0.7629	0.0005	0.1985	0.0591	12.0968
0.5533	11.0	4125	0.7826	0.0027	0.2126	0.0631	16.9146
0.5279	12.0	4500	0.7907	0.0025	0.2144	0.0626	16.656
0.5279	13.0	4875	0.7827	0.0007	0.2019	0.0606	12.4734
0.4964	14.0	5250	0.7933	0.0023	0.2204	0.0674	15.344
0.4803	15.0	5625	0.8169	0.0012	0.1986	0.0594	14.004

Framework versions

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