mistral-qlora_v6

This model is a fine-tuned version of mistralai/Mistral-7B-v0.3 on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 1.2442
• Mean Token Accuracy: 0.6935
• Num Tokens: 896294

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Token Accuracy	Num Tokens
1.3942	0.5161	16	1.3179	0.6832	896294
1.2933	1.0323	32	1.2664	0.6900	896294
1.2234	1.5484	48	1.2503	0.6927	896294
1.2371	2.0645	64	1.2438	0.6937	896294
1.2157	2.5806	80	1.2438	0.6935	896294
1.1698	3.0968	96	1.2439	0.6936	896294
1.1822	3.6129	112	1.2442	0.6935	896294

Framework versions

• PEFT 0.12.0
• Transformers 4.44.2
• Pytorch 2.8.0+cu126
• Datasets 2.20.0
• Tokenizers 0.19.1

Best checkpoint: checkpoint-64