Change a parameter name in FlaxBartForConditionalGeneration.decode() (#13074)

* Change FlaxBartForConditionalGeneration.decode() argument: deterministic -> train

* Also change the parameter name to train for flax marian and mbart
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

src/transformers/models/bart/modeling_flax_bart.py

@@ -1335,7 +1335,7 @@ class FlaxBartForConditionalGeneration(FlaxBartPreTrainedModel):
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
         return_dict: Optional[bool] = None,
-        deterministic: bool = True,
+        train: bool = False,
         params: dict = None,
         dropout_rng: PRNGKey = None,
     ):
@@ -1427,7 +1427,7 @@ class FlaxBartForConditionalGeneration(FlaxBartPreTrainedModel):
             output_attentions=output_attentions,
             output_hidden_states=output_hidden_states,
             return_dict=return_dict,
-            deterministic=deterministic,
+            deterministic=not train,
             rngs=rngs,
             mutable=mutable,
             method=_decoder_forward,

src/transformers/models/marian/modeling_flax_marian.py

@@ -1293,7 +1293,7 @@ class FlaxMarianMTModel(FlaxMarianPreTrainedModel):
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
         return_dict: Optional[bool] = None,
-        deterministic: bool = True,
+        train: bool = False,
         params: dict = None,
         dropout_rng: PRNGKey = None,
     ):
@@ -1385,7 +1385,7 @@ class FlaxMarianMTModel(FlaxMarianPreTrainedModel):
             output_attentions=output_attentions,
             output_hidden_states=output_hidden_states,
             return_dict=return_dict,
-            deterministic=deterministic,
+            deterministic=not train,
             rngs=rngs,
             mutable=mutable,
             method=_decoder_forward,

src/transformers/models/mbart/modeling_flax_mbart.py

@@ -1359,7 +1359,7 @@ class FlaxMBartForConditionalGeneration(FlaxMBartPreTrainedModel):
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
         return_dict: Optional[bool] = None,
-        deterministic: bool = True,
+        train: bool = False,
         params: dict = None,
         dropout_rng: PRNGKey = None,
     ):
@@ -1451,7 +1451,7 @@ class FlaxMBartForConditionalGeneration(FlaxMBartPreTrainedModel):
             output_attentions=output_attentions,
             output_hidden_states=output_hidden_states,
             return_dict=return_dict,
-            deterministic=deterministic,
+            deterministic=not train,
             rngs=rngs,
             mutable=mutable,
             method=_decoder_forward,