train_mrpc_1752826679

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: 0.1926
• 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
7.0586	0.5	413	6.9894	170336
2.6606	1.0	826	2.5820	339568
1.3525	1.5	1239	1.0737	509456
0.5717	2.0	1652	0.4884	678688
0.3105	2.5	2065	0.3018	847104
0.1992	3.0	2478	0.2494	1017368
0.2427	3.5	2891	0.2280	1187704
0.2324	4.0	3304	0.2163	1356744
0.1545	4.5	3717	0.2090	1526088
0.1736	5.0	4130	0.2052	1694912
0.2032	5.5	4543	0.1980	1864000
0.1573	6.0	4956	0.1991	2033992
0.2011	6.5	5369	0.1943	2203208
0.1563	7.0	5782	0.1941	2372464
0.1754	7.5	6195	0.1946	2540880
0.1757	8.0	6608	0.1938	2710624
0.1737	8.5	7021	0.1940	2880128
0.2554	9.0	7434	0.1926	3049392
0.2058	9.5	7847	0.1933	3218352
0.1835	10.0	8260	0.1927	3388032

Framework versions

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