update model card README.md

README.md

@@ -99,10 +99,10 @@ The following hyperparameters were used during training:
                                              'fine_prefix': '<|fine|>',
                                              'misaligned_prefix': '<|misaligned|>',
                                              'substandard_prefix': '<|substandard|>',
-                                             'threshold1': 0.0006421,
-                                             'threshold2': 0.0007833,
-                                             'threshold3': 0.001382,
-                                             'threshold4': 0.999},
+                                             'threshold1': 0.00064215,
+                                             'threshold2': 0.00078331,
+                                             'threshold3': 0.00138205,
+                                             'threshold4': 0.9992},
              'datasets': ['tomekkorbak/detoxify-pile-chunk3-0-50000',
                           'tomekkorbak/detoxify-pile-chunk3-50000-100000',
                           'tomekkorbak/detoxify-pile-chunk3-100000-150000',
@@ -143,7 +143,7 @@ The following hyperparameters were used during training:
                           'tomekkorbak/detoxify-pile-chunk3-1850000-1900000',
                           'tomekkorbak/detoxify-pile-chunk3-1900000-1950000'],
              'is_split_by_sentences': True},
- 'generation': {'force_call_on': [25354],
+ 'generation': {'force_call_on': [22888],
                 'metrics_configs': [{}, {'n': 1}, {'n': 2}, {'n': 5}],
                 'scenario_configs': [{'generate_kwargs': {'bad_words_ids': [[50257],
                                                                             [50258],
@@ -174,7 +174,7 @@ The following hyperparameters were used during training:
                                       'prompt_before_control': True,
                                       'prompts_path': 'resources/challenging_rtp.jsonl'}],
                 'scorer_config': {'device': 'cuda:0'}},
- 'kl_gpt3_callback': {'force_call_on': [25354],
+ 'kl_gpt3_callback': {'force_call_on': [22888],
                       'gpt3_kwargs': {'model_name': 'davinci'},
                       'max_tokens': 64,
                       'num_samples': 4096,
@@ -212,4 +212,4 @@ The following hyperparameters were used during training:
               'weight_decay': 0.1}}
 
 # Wandb URL:
-https://wandb.ai/kejian/uncategorized/runs/3kooiciy
+https://wandb.ai/kejian/uncategorized/runs/20zd4b2c