py38-pylingual-v1-statement

This model is a fine-tuned version of Salesforce/codet5-base on the syssec-utd/statement-py38-pylingual-v1-tokenized dataset.

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: 2e-05
• train_batch_size: 24
• eval_batch_size: 24
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 48
• total_eval_batch_size: 48
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Framework versions

• Transformers 4.48.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.21.0