train_mrpc_1754652142

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: 0.1395
• 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
0.2989	0.5	413	0.1758	170336
0.1096	1.0	826	0.1457	339568
0.161	1.5	1239	0.1939	509456
0.0214	2.0	1652	0.1303	678688
0.1076	2.5	2065	0.1298	847104
0.0259	3.0	2478	0.1248	1017368
0.1086	3.5	2891	0.1059	1187704
0.1231	4.0	3304	0.1156	1356744
0.0881	4.5	3717	0.1146	1526088
0.1287	5.0	4130	0.1349	1694912
0.259	5.5	4543	0.1434	1864000
0.0429	6.0	4956	0.1030	2033992
0.1231	6.5	5369	0.1323	2203208
0.0422	7.0	5782	0.1177	2372464
0.1697	7.5	6195	0.1324	2540880
0.11	8.0	6608	0.1313	2710624
0.0609	8.5	7021	0.1579	2880128
0.1143	9.0	7434	0.1487	3049392
0.0957	9.5	7847	0.1515	3218352
0.0692	10.0	8260	0.1520	3388032

Framework versions

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