mistral-7b-magyar-portas

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1194

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.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: 10
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.1969	0.2195	25	0.1800
0.166	0.4391	50	0.1447
0.1384	0.6586	75	0.1202
0.1691	0.8782	100	0.1142
0.104	1.0966	125	0.1117
0.1048	1.3161	150	0.1134
0.1055	1.5357	175	0.1142
0.0923	1.7552	200	0.1151
0.0877	1.9748	225	0.1129
0.0525	2.1932	250	0.1184
0.0728	2.4127	275	0.1182
0.0565	2.6323	300	0.1195
0.0863	2.8518	325	0.1194

Framework versions

• PEFT 0.15.2
• Transformers 4.52.2
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1