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: 1.4051

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: Use paged_adamw_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
4.5911	0.9231	3	3.9578
4.0234	1.8462	6	3.3966
3.3939	2.7692	9	2.8839
2.142	4.0	13	2.3749
2.4577	4.9231	16	2.1480
2.101	5.8462	19	1.8464
1.7438	6.7692	22	1.6293
1.1545	8.0	26	1.4334
1.3997	8.9231	29	1.3780
1.3138	9.8462	32	1.3435
1.2874	10.7692	35	1.3251
0.9032	12.0	39	1.3148
1.1921	12.9231	42	1.3117
1.1328	13.8462	45	1.3106
1.1094	14.7692	48	1.3166
0.8124	16.0	52	1.3118
1.0274	16.9231	55	1.3233
0.9873	17.8462	58	1.3245
0.9784	18.7692	61	1.3286
0.7057	20.0	65	1.3537
0.9378	20.9231	68	1.3501
0.8876	21.8462	71	1.3664
0.8771	22.7692	74	1.3828
0.658	24.0	78	1.3746
0.8626	24.9231	81	1.3890
0.8507	25.8462	84	1.4078
0.8198	26.7692	87	1.4084
0.5458	27.6923	90	1.4051

Framework versions

• PEFT 0.13.2
• Transformers 4.46.2
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3