train_svamp_1757340271

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: 0.1925
• Num Input Tokens Seen: 704272

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: 101112
• 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
1.9771	0.5	79	1.8618	35296
0.1001	1.0	158	0.2903	70400
0.1223	1.5	237	0.2411	106208
0.264	2.0	316	0.2179	140736
0.0979	2.5	395	0.2146	176064
0.0665	3.0	474	0.2049	211024
0.0938	3.5	553	0.1925	246128
0.1502	4.0	632	0.2071	281616
0.1901	4.5	711	0.2056	316976
0.0696	5.0	790	0.2072	352256
0.0754	5.5	869	0.2153	387360
0.0404	6.0	948	0.1997	422464
0.0762	6.5	1027	0.2075	457760
0.0235	7.0	1106	0.2102	492912
0.0563	7.5	1185	0.2113	528336
0.0215	8.0	1264	0.2108	563600
0.028	8.5	1343	0.2107	598992
0.0469	9.0	1422	0.2096	633984
0.0429	9.5	1501	0.2094	669152
0.0317	10.0	1580	0.2079	704272

Framework versions

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