custom-flan-t5-small-hallucination-classification

This model is a fine-tuned version of google/flan-t5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6832
• Precision: 0.7473
• Recall: 0.7470
• F1: 0.7393
• Accuracy: 0.7470

Model description

Classification head of T5 changed. Layer normalization and GELU added.

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.9007	0.4016	100	0.8154	0.6926	0.6596	0.6202	0.6596
0.7915	0.8032	200	0.7602	0.7181	0.6867	0.6585	0.6867
0.7089	1.2048	300	0.6769	0.7323	0.7329	0.7250	0.7329
0.6381	1.6064	400	0.6954	0.7338	0.7329	0.7240	0.7329
0.6528	2.0080	500	0.6510	0.7423	0.7410	0.7318	0.7410
0.5652	2.4096	600	0.6856	0.7340	0.7339	0.7303	0.7339
0.5446	2.8112	700	0.6832	0.7473	0.7470	0.7393	0.7470

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1