shawgpt-ft

This model is a fine-tuned version of TheBloke/Mistral-7B-Instruct-v0.2-GPTQ on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4144

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
3.9264	0.9231	3	3.3245
3.3146	1.8462	6	2.7427
2.7192	2.7692	9	2.3390
1.7666	4.0	13	2.0350
2.062	4.9231	16	1.8879
1.8845	5.8462	19	1.7629
1.7761	6.7692	22	1.7075
1.2852	8.0	26	1.6366
1.6574	8.9231	29	1.5764
1.5831	9.8462	32	1.5635
1.5859	10.7692	35	1.5347
1.1421	12.0	39	1.5025
1.5139	12.9231	42	1.4772
1.4591	13.8462	45	1.4619
1.4615	14.7692	48	1.4502
1.0953	16.0	52	1.4402
1.4286	16.9231	55	1.4256
1.402	17.8462	58	1.4171
0.9867	18.4615	60	1.4144

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.1.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1