train_mrpc_1753094150

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: 0.1392
• 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
0.1753	0.5	413	0.2030	170336
0.1367	1.0	826	0.1742	339568
0.1458	1.5	1239	0.1701	509456
0.0864	2.0	1652	0.1559	678688
0.1488	2.5	2065	0.1526	847104
0.0996	3.0	2478	0.1551	1017368
0.1873	3.5	2891	0.1515	1187704
0.1651	4.0	3304	0.1471	1356744
0.1176	4.5	3717	0.1468	1526088
0.1225	5.0	4130	0.1450	1694912
0.1569	5.5	4543	0.1424	1864000
0.0896	6.0	4956	0.1408	2033992
0.0668	6.5	5369	0.1403	2203208
0.0834	7.0	5782	0.1402	2372464
0.2288	7.5	6195	0.1406	2540880
0.0935	8.0	6608	0.1395	2710624
0.081	8.5	7021	0.1392	2880128
0.1395	9.0	7434	0.1397	3049392
0.0927	9.5	7847	0.1400	3218352
0.1098	10.0	8260	0.1399	3388032

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1