codet5p-770m-py-codebleu-32-True-1e-06-0.1

This model is a fine-tuned version of Salesforce/codet5p-770m-py on the mbpp dataset. It achieves the following results on the evaluation set:

• Loss: 0.8087
• Codebleu: 0.0867
• Ngram Match Score: 0.0137
• Weighted Ngram Match Score: 0.0422
• Syntax Match Score: 0.1204
• Dataflow Match Score: 0.0824

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: 1e-06
• train_batch_size: 6
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 192
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Codebleu	Ngram Match Score	Weighted Ngram Match Score	Syntax Match Score	Dataflow Match Score
1.9228	0.51	1	0.9113	0.0047	0.0000	0.0000	0.0048	0.0070
0.9857	1.52	3	0.9112	0.0047	0.0000	0.0000	0.0048	0.0070
0.9734	2.54	5	0.9112	0.0069	0.0000	0.0001	0.0067	0.0105
0.9624	3.56	7	0.9111	0.0074	0.0000	0.0002	0.0072	0.0112
0.9586	4.57	9	0.9107	0.0087	0.0000	0.0003	0.0092	0.0126
0.9708	5.59	11	0.9097	0.0140	0.0000	0.0019	0.0178	0.0168
0.9667	6.6	13	0.9092	0.0171	0.0000	0.0034	0.0202	0.0216
0.9791	7.62	15	0.9058	0.0211	0.0000	0.0057	0.0255	0.0258
0.9702	8.63	17	0.9048	0.0317	0.0001	0.0144	0.0366	0.0391
0.9563	9.65	19	0.9034	0.0398	0.0007	0.0192	0.0477	0.0468
0.9654	10.67	21	0.8927	0.0482	0.0014	0.0215	0.0583	0.0566
0.9458	11.68	23	0.8898	0.0602	0.0043	0.0275	0.0742	0.0684
0.9523	12.7	25	0.8866	0.0647	0.0053	0.0286	0.0829	0.0705
0.942	13.71	27	0.8847	0.0786	0.0091	0.0338	0.1069	0.0789
0.94	14.73	29	0.8648	0.0798	0.0099	0.0357	0.1079	0.0803
0.9025	15.75	31	0.8604	0.0809	0.0105	0.0363	0.1122	0.0782
0.9058	16.76	33	0.8577	0.0815	0.0107	0.0362	0.1132	0.0789
0.893	17.78	35	0.8543	0.0816	0.0110	0.0363	0.1132	0.0789
0.8959	18.79	37	0.8524	0.0805	0.0109	0.0362	0.1113	0.0782
0.877	19.81	39	0.8422	0.0808	0.0118	0.0385	0.1113	0.0782
0.861	20.83	41	0.8374	0.0811	0.0118	0.0385	0.1113	0.0789
0.8365	21.84	43	0.8376	0.0827	0.0119	0.0386	0.1132	0.0810
0.8293	22.86	45	0.8331	0.0853	0.0126	0.0390	0.1180	0.0824
0.8288	23.87	47	0.8246	0.0852	0.0134	0.0421	0.1180	0.0810
0.8175	24.89	49	0.8141	0.0852	0.0134	0.0421	0.1180	0.0810
0.6345	25.4	50	0.8087	0.0867	0.0137	0.0422	0.1204	0.0824

Framework versions

• Transformers 4.30.0.dev0
• Pytorch 2.0.1
• Datasets 2.13.1
• Tokenizers 0.13.3