train_svamp_101112_1760638001

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: 0.2259
• Num Input Tokens Seen: 1430592

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: 0.001
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.2239	1.0	158	0.2943	71552
0.246	2.0	316	0.2062	142960
0.0528	3.0	474	0.1448	214432
0.0997	4.0	632	0.1533	286000
0.0227	5.0	790	0.1559	357888
0.0387	6.0	948	0.1550	429456
0.0222	7.0	1106	0.1580	501136
0.0233	8.0	1264	0.1943	573104
0.0281	9.0	1422	0.2004	644752
0.036	10.0	1580	0.1924	716192
0.0096	11.0	1738	0.2382	787200
0.0017	12.0	1896	0.2249	858736
0.001	13.0	2054	0.2267	930160
0.0012	14.0	2212	0.2471	1001792
0.0017	15.0	2370	0.2629	1073248
0.0002	16.0	2528	0.2719	1144672
0.0003	17.0	2686	0.2751	1216160
0.0003	18.0	2844	0.2769	1287728
0.0005	19.0	3002	0.2809	1359120
0.0001	20.0	3160	0.2799	1430592

Framework versions

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