TinyMistral-specific-4bit

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: 1.1940

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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.7744	0.8696	5	1.5917
1.3359	1.9130	11	1.4566
1.254	2.9565	17	1.3934
1.1805	4.0	23	1.3203
1.3421	4.8696	28	1.2766
1.0701	5.9130	34	1.2354
1.0341	6.9565	40	1.2107
1.0113	8.0	46	1.1974
1.0482	8.6957	50	1.1940

Framework versions

• PEFT 0.13.2
• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.20.3