train_conala_1754507517

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

• Loss: 0.6303
• Num Input Tokens Seen: 1524216

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
1.8549	0.5	268	1.8534	75936
1.3657	1.0	536	0.9853	152672
0.7804	1.5	804	0.8251	229344
0.8644	2.0	1072	0.7605	305288
0.6738	2.5	1340	0.7196	382120
0.5981	3.0	1608	0.6914	457952
0.6263	3.5	1876	0.6750	534688
0.4265	4.0	2144	0.6613	610944
0.526	4.5	2412	0.6523	687328
0.953	5.0	2680	0.6465	762440
0.6426	5.5	2948	0.6416	839656
0.6051	6.0	3216	0.6378	914920
0.5827	6.5	3484	0.6353	992104
0.4853	7.0	3752	0.6351	1067520
0.733	7.5	4020	0.6322	1142912
0.6035	8.0	4288	0.6314	1220200
0.7938	8.5	4556	0.6307	1295720
0.765	9.0	4824	0.6308	1372560
0.6362	9.5	5092	0.6303	1447376
0.4372	10.0	5360	0.6308	1524216

Framework versions

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