0.1.0

First version of the model, results:

-- Token-Level Evaluation ----

Detailed Classification Report:
              precision    recall  f1-score   support

   Supported     0.9799    0.9859    0.9829    422046
Hallucinated     0.6096    0.5222    0.5625     17844

    accuracy                         0.9671    439890
   macro avg     0.7947    0.7540    0.7727    439890
weighted avg     0.9649    0.9671    0.9658    439890

Evaluation Results:

Hallucination Detection (Class 1):
  Precision: 0.6096
  Recall: 0.5222
  F1: 0.5625

Supported Content (Class 0):
  Precision: 0.9799
  Recall: 0.9859
  F1: 0.9829

---- Example-Level Evaluation ----

Detailed Example-Level Classification Report:
              precision    recall  f1-score   support

   Supported     0.8696    0.8765    0.8730      1757
Hallucinated     0.7664    0.7550    0.7607       943

    accuracy                         0.8341      2700
   macro avg     0.8180    0.8158    0.8168      2700
weighted avg     0.8335    0.8341    0.8338      2700

Example-Level Evaluation Results:

Hallucination Detection (Example Level) - Class 1:
  Precision: 0.7664
  Recall: 0.7550
  F1: 0.7607

Supported Content (Example Level) - Class 0:
  Precision: 0.8696
  Recall: 0.8765
  F1: 0.8730

Full Changelog: https://github.com/KRLabsOrg/LettuceDetect/commits/0.1.0