codellama_instruct_spider_e10

This model is a fine-tuned version of codellama/CodeLlama-7b-Instruct-hf on the tmnam20/SpiderInstruct dataset. It achieves the following results on the evaluation set:

• Loss: 0.2393

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.822	0.37	100	0.5313
0.3014	0.74	200	0.2763
0.2091	1.11	300	0.2469
0.1697	1.48	400	0.2401
0.1495	1.85	500	0.2395
0.1256	2.22	600	0.2525
0.1097	2.59	700	0.2641
0.1107	2.96	800	0.2617
0.0951	3.33	900	0.2683
0.0882	3.7	1000	0.2892
0.0818	4.06	1100	0.3134
0.075	4.43	1200	0.2978
0.0745	4.8	1300	0.3095
0.0642	5.17	1400	0.3261
0.0622	5.54	1500	0.3201
0.0573	5.91	1600	0.3343
0.0552	6.28	1700	0.3396
0.0523	6.65	1800	0.3602
0.0538	7.02	1900	0.3464
0.0467	7.39	2000	0.3622
0.0465	7.76	2100	0.3697
0.044	8.13	2200	0.3890
0.043	8.5	2300	0.3785
0.0375	8.87	2400	0.3860
0.0384	9.24	2500	0.3952
0.0363	9.61	2600	0.3940
0.0352	9.98	2700	0.3985

Framework versions

• Transformers 4.34.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3