train_record_42_1760886907

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the record dataset. It achieves the following results on the evaluation set:

• Loss: 0.7495
• Num Input Tokens Seen: 825879088

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.3846	2.0	55542	0.3528	82612264
0.3616	4.0	111084	0.2995	165185880
0.1859	6.0	166626	0.3165	247772264
0.3768	8.0	222168	0.3529	330363216
0.1902	10.0	277710	0.3780	412933248
0.1147	12.0	333252	0.4336	495530312
0.0946	14.0	388794	0.4956	578135608
0.0762	16.0	444336	0.5760	660689216
0.1649	18.0	499878	0.6866	743284656
0.0681	20.0	555420	0.7495	825879088

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1