train_svamp_1757340197

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: 0.0973
• Num Input Tokens Seen: 705184

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
1.8026	0.5	79	1.4394	35776
0.117	1.0	158	0.1366	70672
0.0931	1.5	237	0.1189	105904
0.1259	2.0	316	0.1163	141328
0.0836	2.5	395	0.1113	176752
0.0589	3.0	474	0.1143	211808
0.0598	3.5	553	0.1081	247104
0.1036	4.0	632	0.1026	282048
0.0738	4.5	711	0.0975	317248
0.1063	5.0	790	0.0975	352592
0.0617	5.5	869	0.0973	388176
0.0871	6.0	948	0.0979	423184
0.0905	6.5	1027	0.1042	458640
0.0633	7.0	1106	0.0979	493440
0.1537	7.5	1185	0.0979	528768
0.047	8.0	1264	0.0990	563872
0.0195	8.5	1343	0.0974	599232
0.1123	9.0	1422	0.0987	634544
0.0886	9.5	1501	0.0992	670064
0.0054	10.0	1580	0.0986	705184

Framework versions

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