train_mrpc_1754502819

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

• Loss: 0.1519
• Num Input Tokens Seen: 3388032

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: 4
• eval_batch_size: 4
• seed: 123
• 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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.1751	0.5	413	0.1808	170336
0.1631	1.0	826	0.1586	339568
0.1683	1.5	1239	0.1765	509456
0.0535	2.0	1652	0.1683	678688
0.1837	2.5	2065	0.1569	847104
0.0397	3.0	2478	0.1648	1017368
0.1598	3.5	2891	0.1597	1187704
0.185	4.0	3304	0.1519	1356744
0.1661	4.5	3717	0.1730	1526088
0.1496	5.0	4130	0.1609	1694912
0.3434	5.5	4543	0.1826	1864000
0.0678	6.0	4956	0.1609	2033992
0.0916	6.5	5369	0.1694	2203208
0.0281	7.0	5782	0.1653	2372464
0.2431	7.5	6195	0.1725	2540880
0.1769	8.0	6608	0.1703	2710624
0.1186	8.5	7021	0.1696	2880128
0.1694	9.0	7434	0.1707	3049392
0.1561	9.5	7847	0.1705	3218352
0.1541	10.0	8260	0.1703	3388032

Framework versions

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