Mistral-7B-text-to-sql-flash-attention-2-FAISS-10epoch

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.4891

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: 0.0002
• train_batch_size: 3
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 24
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.03
• lr_scheduler_warmup_steps: 15
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
2.1635	0.2005	10	0.7133
0.6525	0.4010	20	0.5602
0.5386	0.6015	30	0.5175
0.5042	0.8020	40	0.4926
0.487	1.0025	50	0.4839
0.4099	1.2030	60	0.4804
0.4078	1.4035	70	0.4748
0.4125	1.6040	80	0.4692
0.402	1.8045	90	0.4645
0.398	2.0050	100	0.4635
0.3066	2.2055	110	0.5003
0.3049	2.4060	120	0.4879
0.313	2.6065	130	0.4891

Framework versions

• PEFT 0.11.1
• Transformers 4.42.3
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1