DREAM-t5-small

This model is a fine-tuned version of t5-small on th DREAM dataset. It achieves the following results on the evaluation set:

• Loss: 1.6873

Model description

The model's expected input is of the form

"[SITUATION] Best friend cheated on his girlfriend with me…. [QUERY] rot"

to which the model replies

"It's wrong to cheat on your partner."

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: 0.0003
• train_batch_size: 32
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.8513	1.0	2447	1.7593
1.7482	2.0	4894	1.7202
1.6892	3.0	7341	1.7004
1.6415	4.0	9788	1.6944
1.6103	5.0	12235	1.6886
1.5856	6.0	14682	1.6873

Framework versions

• Transformers 4.35.2
• Pytorch 2.0.0+cu117
• Datasets 2.15.0
• Tokenizers 0.15.0