train_conala_42_1760637551

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: 2.8491
• Num Input Tokens Seen: 3049984

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: 42
• 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
3.3068	1.0	536	2.8759	153352
2.4802	2.0	1072	2.8654	305496
2.1282	3.0	1608	2.8566	458160
2.6148	4.0	2144	2.8538	610584
2.8073	5.0	2680	2.8527	763216
2.8808	6.0	3216	2.8518	915528
2.9918	7.0	3752	2.8497	1067904
2.4865	8.0	4288	2.8512	1221016
3.2252	9.0	4824	2.8516	1373032
4.3428	10.0	5360	2.8491	1525104
2.8065	11.0	5896	2.8523	1677680
3.4089	12.0	6432	2.8510	1830200
2.869	13.0	6968	2.8503	1982664
2.7414	14.0	7504	2.8524	2135168
2.9479	15.0	8040	2.8501	2287232
3.1305	16.0	8576	2.8506	2438992
2.6897	17.0	9112	2.8494	2591432
3.2659	18.0	9648	2.8503	2744944
2.6155	19.0	10184	2.8508	2897552
3.0457	20.0	10720	2.8508	3049984

Framework versions

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