llama3-8B_less_data_for_fact_update

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: 0.7586
• Accuracy: 0.7733
• Precision: 0.8015
• Recall: 0.7267
• F1 score: 0.7622

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
0.8492	0.5	200	0.8538	0.65	0.6923	0.54	0.6067
0.7574	1.0	400	0.6427	0.7267	0.8036	0.6	0.6870
0.5003	1.5	600	0.5917	0.73	0.7066	0.7867	0.7445
0.4629	2.0	800	0.6470	0.7567	0.8235	0.6533	0.7286
0.3114	2.5	1000	0.6980	0.6967	0.7438	0.6	0.6642
0.3031	3.0	1200	0.6128	0.7567	0.8130	0.6667	0.7326
0.1783	3.5	1400	0.7288	0.7667	0.8175	0.6867	0.7464
0.1737	4.0	1600	0.7242	0.7533	0.7969	0.68	0.7338
0.0924	4.5	1800	0.7214	0.7733	0.7808	0.76	0.7703
0.0669	5.0	2000	0.7586	0.7733	0.8015	0.7267	0.7622

Framework versions

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