metadata

library_name: transformers
license: apache-2.0
base_model: Salesforce/codet5-base
tags:
  - generated_from_trainer
model-index:
  - name: py38-pylingual-v1.1.1-statement
    results: []

py38-pylingual-v1.1.1-statement

This model is a fine-tuned version of Salesforce/codet5-base on the syssec-utd/statement-py38-pylingual-v1.1.1-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: 3
total_train_batch_size: 72
total_eval_batch_size: 72
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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 4.54.1
Pytorch 2.8.0+cu128
Datasets 3.3.2
Tokenizers 0.21.4