shawgpt-ft-epoch20

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: 1.3738

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
4.6183 0.9231 3 4.0898
4.2697 1.8462 6 3.7187
3.8116 2.7692 9 3.3358
2.5344 4.0 13 2.8801
3.0199 4.9231 16 2.6124
2.6926 5.8462 19 2.3845
2.4042 6.7692 22 2.1734
1.6092 8.0 26 1.8884
1.8915 8.9231 29 1.7507
1.7133 9.8462 32 1.6384
1.607 10.7692 35 1.5362
1.1015 12.0 39 1.4586
1.4428 12.9231 42 1.4286
1.3842 13.8462 45 1.4070
1.3787 14.7692 48 1.3934
1.0289 16.0 52 1.3819
1.3426 16.9231 55 1.3771
1.3259 17.8462 58 1.3744
0.935 18.4615 60 1.3738

Framework versions

  • PEFT 0.13.2
  • Transformers 4.44.2
  • Pytorch 2.5.0+cu124
  • Datasets 3.0.1
  • Tokenizers 0.19.1
Downloads last month
0
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for FrederikKlinkby/shawgpt-ft-epoch20