train_conala_1754652181

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: 1.8173
• 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
0.7295	0.5	268	0.7194	75936
0.9109	1.0	536	0.6632	152672
0.6458	1.5	804	0.6389	229344
0.6543	2.0	1072	0.6025	305288
0.6009	2.5	1340	0.5999	382120
0.4877	3.0	1608	0.5991	457952
0.5311	3.5	1876	0.5987	534688
0.3989	4.0	2144	0.6024	610944
0.4024	4.5	2412	0.6075	687328
0.8237	5.0	2680	0.5997	762440
0.4437	5.5	2948	0.6065	839656
0.5024	6.0	3216	0.6062	914920
0.421	6.5	3484	0.6185	992104
0.4054	7.0	3752	0.6159	1067520
0.5722	7.5	4020	0.6261	1142912
0.4968	8.0	4288	0.6193	1220200
0.6181	8.5	4556	0.6358	1295720
0.6561	9.0	4824	0.6277	1372560
0.4826	9.5	5092	0.6322	1447376
0.2695	10.0	5360	0.6315	1524216

Framework versions

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