train_svamp_456_1757596108

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

• Loss: 0.1483
• Num Input Tokens Seen: 1352544

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: 2
• eval_batch_size: 2
• seed: 456
• 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
0.5567	1.0	315	0.6299	67504
0.2481	2.0	630	0.2979	135312
0.0177	3.0	945	0.1330	202928
0.0472	4.0	1260	0.0844	270592
0.022	5.0	1575	0.0908	338048
0.0583	6.0	1890	0.0926	405584
0.0008	7.0	2205	0.1284	473216
0.0184	8.0	2520	0.1032	541040
0.0002	9.0	2835	0.1117	608592
0.0	10.0	3150	0.1170	676256
0.0	11.0	3465	0.1389	743840
0.0	12.0	3780	0.1512	811616
0.0	13.0	4095	0.1398	879184
0.0001	14.0	4410	0.1442	947024
0.0	15.0	4725	0.1473	1014704
0.0	16.0	5040	0.1476	1082128
0.0	17.0	5355	0.1478	1149696
0.0	18.0	5670	0.1479	1217440
0.0	19.0	5985	0.1486	1284960
0.0	20.0	6300	0.1483	1352544

Framework versions

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