Upload benchmark_results.md with huggingface_hub

benchmark_results.md

@@ -8,7 +8,7 @@ The best performing model was chosen based on the highest average F1 score acros
 ```
 learning_rate: 5e-05
 per_device_train_batch_size: 32
-num_train_epochs: 8
+num_train_epochs: 10
 weight_decay: 0.05
 lr_scheduler_type: cosine
 ```
@@ -17,8 +17,8 @@ lr_scheduler_type: cosine
 
 | dataset                           |   accuracy |   f1_score |   recall |   precision |
 |:----------------------------------|-----------:|-----------:|---------:|------------:|
-| BRlkl/BingoGuard-train-test-pt    |   0.88664  |   0.939914 | 0.88664  |    1        |
-| BRlkl/openai-moderation-eval-pt   |   0.715476 |   0.661473 | 0.894636 |    0.524719 |
-| BRlkl/WildGuardTest-pt            |   0.845203 |   0.822657 | 0.809019 |    0.836763 |
-| BRlkl/XSTest-pt                   |   0.828889 |   0.821346 | 0.885    |    0.766234 |
-| BRlkl/toxic-chat-pt (40% holdout) |   0.976401 |   0.854545 | 0.927632 |    0.792135 |
+| BRlkl/BingoGuard-train-test-pt    |   0.868421 |   0.929577 | 0.868421 |    1        |
+| BRlkl/openai-moderation-eval-pt   |   0.71131  |   0.655784 | 0.885057 |    0.520857 |
+| BRlkl/WildGuardTest-pt            |   0.837552 |   0.803698 | 0.749337 |    0.866564 |
+| BRlkl/XSTest-pt                   |   0.877778 |   0.864865 | 0.88     |    0.850242 |
+| BRlkl/toxic-chat-pt (40% holdout) |   0.974435 |   0.843373 | 0.921053 |    0.777778 |