logdataset1024samples15epochs_llama3-1_16bit_64eval_natural_adalora_2

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B on the svenwey/LogDataset dataset. It achieves the following results on the evaluation set:

• Loss: 3.0752
• Model Preparation Time: 0.0137
• Linecount Difference Smape Score: 0.7773
• Linecontentlength Difference Smape Score: 0.6338
• Linecontent Sacrebleu Score: 0.4516
• Linecontent Sacrebleu Withoutexplicitnumbers Score: 0.4777
• Timestamps Smape Difference Score: 0.7332
• Timestamps Formatconsistency Score: 1.0
• Timestamps Monotinicity Score: 1.0

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: 5e-05
• train_batch_size: 1
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Linecount Difference Smape Score	Linecontentlength Difference Smape Score	Linecontent Sacrebleu Score	Linecontent Sacrebleu Withoutexplicitnumbers Score	Timestamps Smape Difference Score	Timestamps Formatconsistency Score	Timestamps Monotinicity Score
1.0869	0.9998	1147	1.1504	0.0137	0.7016	0.5256	0.3531	0.3637	0.6414	0.9531	0.9531
0.6665	1.9996	2294	1.5606	0.0137	0.7120	0.5669	0.4394	0.4462	0.6706	0.9844	0.9844
0.7854	2.9993	3441	1.7849	0.0137	0.7278	0.5911	0.4532	0.4643	0.6926	0.9688	0.9688
0.9092	4.0	4589	1.9795	0.0137	0.7370	0.5918	0.4415	0.4605	0.6926	0.9688	0.9688
1.0273	4.9998	5736	2.1496	0.0137	0.7426	0.5876	0.4323	0.4541	0.6975	0.9844	0.9844
1.1379	5.9996	6883	2.3043	0.0137	0.7554	0.6070	0.4534	0.4805	0.7112	0.9844	0.9844
1.2381	6.9993	8030	2.4486	0.0137	0.7618	0.6078	0.4553	0.4755	0.7194	0.9844	0.9844
1.326	8.0	9178	2.5735	0.0137	0.7655	0.6224	0.4573	0.4797	0.7235	1.0	1.0
1.4048	8.9998	10325	2.7091	0.0137	0.7602	0.6199	0.4522	0.4765	0.7160	1.0	1.0
1.4702	9.9996	11472	2.8050	0.0137	0.7396	0.5926	0.4319	0.4448	0.6944	0.9844	0.9844
1.5236	10.9993	12619	2.9013	0.0137	0.7473	0.6007	0.4234	0.4437	0.6984	1.0	1.0
1.5639	12.0	13767	2.9654	0.0137	0.7528	0.6115	0.4416	0.4739	0.7073	1.0	1.0
1.5956	12.9998	14914	3.0193	0.0137	0.7666	0.6197	0.4456	0.4609	0.7295	1.0	1.0
1.6152	13.9996	16061	3.0513	0.0137	0.7411	0.5917	0.4208	0.4372	0.6896	0.9844	0.9844
1.6235	14.9967	17205	3.0752	0.0137	0.7773	0.6338	0.4516	0.4777	0.7332	1.0	1.0

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1