train_codealpacapy_101112_1770403545

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

• Loss: 0.6078
• Num Input Tokens Seen: 22265472

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: 1e-05
• 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
0.5307	2.0	3392	0.5230	2224576
0.456	4.0	6784	0.5043	4459088
0.7937	6.0	10176	0.5033	6685232
0.3851	8.0	13568	0.5102	8907352
0.415	10.0	16960	0.5237	11130848
0.3657	12.0	20352	0.5405	13367408
0.2972	14.0	23744	0.5666	15590856
0.3497	16.0	27136	0.5873	17815160
0.3135	18.0	30528	0.6031	20038104
0.3684	20.0	33920	0.6078	22265472

Framework versions

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