shawgpt-ft-model2

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: 1.3687

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
4.6185	0.9231	3	4.0890
4.2705	1.8462	6	3.7193
3.8202	2.7692	9	3.3399
2.543	4.0	13	2.8865
3.0261	4.9231	16	2.6014
2.67	5.8462	19	2.3431
2.3646	6.7692	22	2.1289
1.5668	8.0	26	1.8519
1.8747	8.9231	29	1.7285
1.7057	9.8462	32	1.6162
1.5973	10.7692	35	1.5193
1.0962	12.0	39	1.4458
1.4377	12.9231	42	1.4184
1.3833	13.8462	45	1.4014
1.3812	14.7692	48	1.3898
1.0309	16.0	52	1.3781
1.3451	16.9231	55	1.3725
1.328	17.8462	58	1.3696
0.9369	18.4615	60	1.3687

Framework versions

• PEFT 0.13.2
• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1