Change a parameter in decode(): deterministic -> train

vit_gpt2/modeling_flax_vit_gpt2_lm.py

@@ -398,21 +398,6 @@ class FlaxViTGPT2LMPreTrainedModel(FlaxPreTrainedModel):
         )
 
 
-# @add_start_docstrings(
-#     "The bare Bart Model transformer outputting raw hidden-states without any specific head on top.",
-#     BART_START_DOCSTRING,
-# )
-# class FlaxViTGPT2LMModel(FlaxViTGPT2LMPreTrainedModel):
-#     config: BartConfig
-#     dtype: jnp.dtype = jnp.float32  # the dtype of the computation
-#     module_class = FlaxViTGPT2LMModule
-#
-#
-# append_call_sample_docstring(
-#     FlaxBartModel, _TOKENIZER_FOR_DOC, _CHECKPOINT_FOR_DOC, FlaxSeq2SeqModelOutput, _CONFIG_FOR_DOC
-# )
-
-
 class FlaxViTGPT2LMForConditionalGeneration(FlaxViTGPT2LMPreTrainedModel):
     module_class = FlaxViTGPT2LMForConditionalGenerationModule
     dtype: jnp.dtype = jnp.float32
@@ -428,7 +413,7 @@ class FlaxViTGPT2LMForConditionalGeneration(FlaxViTGPT2LMPreTrainedModel):
         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,
     ):
@@ -443,7 +428,7 @@ class FlaxViTGPT2LMForConditionalGeneration(FlaxViTGPT2LMPreTrainedModel):
             output_attentions,
             output_hidden_states,
             return_dict,
-            not deterministic,
+            train,
             params,
             dropout_rng,
         )