shawgpt-ft3

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: 4.0882

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: 2e-05
• train_batch_size: 42
• eval_batch_size: 42
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 168
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.1584	1.0	1	4.2318
2.0801	2.0	2	4.2260
2.1425	3.0	3	4.2076
2.1384	4.0	4	4.1900
2.1016	5.0	5	4.1740
2.153	6.0	6	4.1593
2.0855	7.0	7	4.1459
2.0668	8.0	8	4.1338
2.0384	9.0	9	4.1226
2.1113	10.0	10	4.1130
2.0445	11.0	11	4.1049
2.0987	12.0	12	4.0984
2.0449	13.0	13	4.0931
2.0419	14.0	14	4.0899
2.0418	15.0	15	4.0882

Framework versions

• PEFT 0.13.2
• Transformers 4.44.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.2
• Tokenizers 0.19.1