train_gsm8k_42_1760637597

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the gsm8k dataset. It achieves the following results on the evaluation set:

• Loss: 0.5047
• Num Input Tokens Seen: 34797032

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: 42
• 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
0.9545	1.0	1682	0.9514	1738904
0.696	2.0	3364	0.6189	3481872
0.6072	3.0	5046	0.5705	5222160
0.4677	4.0	6728	0.5498	6964040
0.5573	5.0	8410	0.5374	8703920
0.4639	6.0	10092	0.5292	10444208
0.5226	7.0	11774	0.5232	12184024
0.6014	8.0	13456	0.5187	13925976
0.5909	9.0	15138	0.5149	15667472
0.5348	10.0	16820	0.5122	17407120
0.4864	11.0	18502	0.5102	19145240
0.4748	12.0	20184	0.5086	20882704
0.5335	13.0	21866	0.5073	22623936
0.525	14.0	23548	0.5063	24361736
0.5153	15.0	25230	0.5057	26106632
0.4356	16.0	26912	0.5052	27845256
0.515	17.0	28594	0.5049	29579704
0.5429	18.0	30276	0.5047	31322600
0.4554	19.0	31958	0.5047	33057576
0.5521	20.0	33640	0.5047	34797032

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.9.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4