shawgpt-ft-model3

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: 1.3192

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: 0.0003
• 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
4.6026	0.9231	3	4.0086
4.0831	1.8462	6	3.4513
3.4469	2.7692	9	2.9279
2.1941	4.0	13	2.4527
2.5155	4.9231	16	2.1409
2.0928	5.8462	19	1.8650
1.7541	6.7692	22	1.6496
1.1763	8.0	26	1.4669
1.4274	8.9231	29	1.4114
1.3494	9.8462	32	1.3828
1.3425	10.7692	35	1.3657
0.9564	12.0	39	1.3474
1.2781	12.9231	42	1.3368
1.2326	13.8462	45	1.3295
1.2348	14.7692	48	1.3254
0.924	16.0	52	1.3217
1.2023	16.9231	55	1.3203
1.1861	17.8462	58	1.3195
0.8402	18.4615	60	1.3192

Framework versions

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