[core] improve API

tests/test_ppo_trainer.py

@@ -98,6 +98,37 @@ def setUp(self):
 
         return super().setUp()
 
+    def test_raise_unvalid_args(self):
+        # initialize dataset
+        dummy_dataset = self._init_dummy_dataset()
+
+        with self.assertRaises(ValueError):
+            _ = PPOTrainer(
+                config={},
+                model=self.gpt2_model,
+                ref_model=self.gpt2_model_ref,
+                tokenizer=self.gpt2_tokenizer,
+                dataset=dummy_dataset,
+            )
+
+        with self.assertRaises(ValueError):
+            _ = PPOTrainer(
+                config=self.ppo_config,
+                model=self.gpt2_model,
+                ref_model=self.gpt2_model_ref,
+                tokenizer=None,
+                dataset=dummy_dataset,
+            )
+
+        with self.assertRaises(ValueError):
+            _ = PPOTrainer(
+                config=self.ppo_config,
+                model=None,
+                ref_model=self.gpt2_model_ref,
+                tokenizer=self.gpt2_tokenizer,
+                dataset=dummy_dataset,
+            )
+
     def test_ppo_step(self):
         # initialize dataset
         dummy_dataset = self._init_dummy_dataset()

trl/trainer/ppo_trainer.py

@@ -65,10 +65,10 @@ class PPOTrainer(BaseTrainer):
 
     def __init__(
         self,
-        config: PPOConfig,
-        model: PreTrainedModelWrapper,
-        ref_model: PreTrainedModelWrapper,
-        tokenizer: Union[PreTrainedTokenizer, PreTrainedTokenizerFast],
+        config: PPOConfig = None,
+        model: PreTrainedModelWrapper = None,
+        ref_model: PreTrainedModelWrapper = None,
+        tokenizer: Union[PreTrainedTokenizer, PreTrainedTokenizerFast] = None,
         dataset: Optional[Union[torch.utils.data.Dataset, Dataset]] = None,
         optimizer: Optional[torch.optim.Optimizer] = None,
         data_collator=None,
@@ -99,16 +99,21 @@ def __init__(
                 used only if `ref_model` is `None`.
         """
         super().__init__(config)
-
-        # Step 1: Initialize Accelerator
-        self.accelerator = Accelerator(log_with=config.log_with, **config.accelerator_kwargs)
-        self.accelerator.init_trackers(config.tracker_project_name, config=config.to_dict(), **config.tracker_kwargs)
-
-        # Step 2: Initialize model, tokenizer, and dataloader
+        # Step 0: check positional arguments validity
+        if not isinstance(config, PPOConfig):
+            raise ValueError(f"config must be a PPOConfig, got {type(config)}")
+        if not isinstance(tokenizer, (PreTrainedTokenizer, PreTrainedTokenizerFast)):
+            raise ValueError(
+                f"tokenizer must be a PreTrainedTokenizer or PreTrainedTokenizerFast, got {type(tokenizer)}"
+            )
         if not isinstance(model, PreTrainedModelWrapper):
             raise ValueError(
                 f"model must be a PreTrainedModelWrapper, got {type(model)} - supported architectures are: {SUPPORTED_ARCHITECTURES}"
             )
+        # Step 1: Initialize Accelerator
+        self.accelerator = Accelerator(log_with=config.log_with, **config.accelerator_kwargs)
+        self.accelerator.init_trackers(config.tracker_project_name, config=config.to_dict(), **config.tracker_kwargs)
+
         self.model = model
         self.is_encoder_decoder = hasattr(self.model, "is_encoder_decoder")