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: 1.7402

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: 8e-05
• 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
4.6249	0.9231	3	4.1254
4.3778	1.8462	6	3.8852
4.0877	2.7692	9	3.6541
2.8518	4.0	13	3.3539
3.5577	4.9231	16	3.1426
3.3003	5.8462	19	2.9494
3.0846	6.7692	22	2.7761
2.1597	8.0	26	2.5710
2.6927	8.9231	29	2.4270
2.5126	9.8462	32	2.3068
2.3827	10.7692	35	2.1855
1.6467	12.0	39	2.0325
2.0858	12.9231	42	1.9558
1.9783	13.8462	45	1.8930
1.9199	14.7692	48	1.8388
1.3996	16.0	52	1.7856
1.8117	16.9231	55	1.7594
1.7748	17.8462	58	1.7441
1.234	18.4615	60	1.7402

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1