llama7bit-lora-sql

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.3666

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 1399
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• training_steps: 500

Training results

Training Loss	Epoch	Step	Validation Loss
1.2114	0.06	20	0.8331
0.6838	0.12	40	0.5232
0.5107	0.17	60	0.4600
0.4618	0.23	80	0.4289
0.4412	0.29	100	0.4153
0.425	0.35	120	0.4067
0.4182	0.41	140	0.3956
0.4137	0.47	160	0.3912
0.4047	0.52	180	0.3865
0.4034	0.58	200	0.3834
0.3968	0.64	220	0.3833
0.3954	0.7	240	0.3782
0.3921	0.76	260	0.3756
0.3877	0.82	280	0.3730
0.3849	0.87	300	0.3722
0.3831	0.93	320	0.3714
0.3787	0.99	340	0.3702
0.3677	1.05	360	0.3692
0.3632	1.11	380	0.3686
0.3611	1.17	400	0.3677
0.3588	1.22	420	0.3669
0.3579	1.28	440	0.3666
0.3551	1.34	460	0.3666
0.3586	1.4	480	0.3665
0.3581	1.46	500	0.3666

Framework versions

• PEFT 0.8.2
• Transformers 4.37.2
• Pytorch 2.2.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.2