llama-7b-offense

This model is a fine-tuned version of meta-llama/Llama-3.1-8B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5788
• Accuracy: 0.7709
• F1: 0.5386

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Use adamw_torch 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: 100
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	0.9662	100	0.7477	0.6895	0.3890
No log	1.9275	200	0.6635	0.7186	0.4739
No log	2.8889	300	0.6167	0.7430	0.5328
No log	3.8502	400	0.6212	0.7384	0.5399
2.5728	4.8116	500	0.5934	0.7558	0.5532
2.5728	5.7729	600	0.5906	0.7605	0.5231
2.5728	6.7343	700	0.5838	0.7698	0.505
2.5728	7.6957	800	0.5780	0.7605	0.5381
2.5728	8.6570	900	0.5799	0.7686	0.5340
1.9058	9.6184	1000	0.5788	0.7709	0.5386

Framework versions

• PEFT 0.14.0
• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0