ad-gen-ft

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

• Loss: 0.0765

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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.5396	0.9455	13	1.0181
0.7048	1.9636	27	0.3883
0.3153	2.9818	41	0.1752
0.1968	4.0	55	0.1220
0.1637	4.9455	68	0.1000
0.1271	5.9636	82	0.0871
0.112	6.9818	96	0.0812
0.1026	8.0	110	0.0782
0.1038	8.9455	123	0.0766
0.0808	9.4545	130	0.0765

Framework versions

• PEFT 0.10.0
• Transformers 4.40.2
• Pytorch 2.1.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1