llama3-8B_Fact_U

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1146
• Accuracy: 0.7953
• Precision: 0.8252
• Recall: 0.7692
• F1 score: 0.7963

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: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1 score
0.9272	0.2509	200	1.1877	0.5765	0.5589	0.8801	0.6837
0.7133	0.5019	400	0.7246	0.7	0.6912	0.7647	0.7261
0.6476	0.7528	600	0.6219	0.7565	0.7945	0.7172	0.7539
0.5766	1.0038	800	0.5714	0.76	0.85	0.6538	0.7391
0.4292	1.2547	1000	0.5870	0.7588	0.7474	0.8100	0.7774
0.4047	1.5056	1200	0.6078	0.7824	0.7694	0.8303	0.7987
0.4753	1.7566	1400	0.5654	0.7929	0.8575	0.7217	0.7838
0.3831	2.0075	1600	0.5672	0.8012	0.8289	0.7783	0.8028
0.2666	2.2585	1800	0.6054	0.8047	0.8416	0.7692	0.8038
0.3267	2.5094	2000	0.6620	0.7929	0.8182	0.7738	0.7953
0.2742	2.7604	2200	0.6985	0.7941	0.8579	0.7240	0.7853
0.3125	3.0113	2400	0.6373	0.7859	0.7955	0.7919	0.7937
0.1743	3.2622	2600	0.8362	0.8047	0.8730	0.7308	0.7956
0.1837	3.5132	2800	0.7889	0.7965	0.8440	0.7466	0.7923
0.2013	3.7641	3000	0.8958	0.7659	0.7425	0.8416	0.7890
0.1572	4.0151	3200	0.8662	0.8012	0.864	0.7330	0.7931
0.0993	4.2660	3400	0.8730	0.8024	0.8477	0.7557	0.7990
0.0704	4.5169	3600	1.0571	0.8012	0.8421	0.7602	0.7990
0.0598	4.7679	3800	1.1146	0.7953	0.8252	0.7692	0.7963

Framework versions

• PEFT 0.11.1
• Transformers 4.44.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1