my_humorous_llm-model

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

my_humorous_llm-model

This model is a fine-tuned version of distilbert/distilgpt2 on the "Reddit-clean-jokes" dataset taken from the following link: https://github.com/amoudgl/short-jokes-dataset/blob/master/data/reddit-cleanjokes.csv. It achieves the following results on the evaluation set:

• Loss: 1.9830

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: 8
• eval_batch_size: 8
• 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: linear
• num_epochs: 3.0

Training results

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3