change to add_cross_attention

vit_gpt2/modeling_flax_gpt2.py

@@ -299,7 +299,7 @@ class FlaxGPT2Block(nn.Module):
 
     def setup(self):
 
-        self.only_self_attn = not self.config.getattr('with_cross_attention', False)
+        self.only_self_attn = not self.config.add_cross_attention
 
         hidden_size = self.config.hidden_size
         inner_dim = self.config.n_inner if self.config.n_inner is not None else 4 * hidden_size
@@ -412,7 +412,7 @@ class FlaxGPT2PreTrainedModel(FlaxPreTrainedModel):
         params_rng, dropout_rng = jax.random.split(rng)
         rngs = {"params": params_rng, "dropout": dropout_rng}
 
-        if self.config.getattr('with_cross_attention', False):
+        if self.config.add_cross_attention:
             encoder_hidden_states = jnp.zeros(input_shape + (self.config.n_embd,))
             encoder_attention_mask = attention_mask
             module_init_outputs = self.module.init(