Fix serving_output for TF composite models (encoder-decoder like models)

[ydshieh]

What does this PR do?

[If the concept is approved, I will apply the same changes to other TF encoder-decoder family of models]

The composite models use its components' configurations. See for example

transformers/src/transformers/modeling_tf_utils.py

Lines 426 to 430 in 95e7057

	# Encoder Decoder models delegate the application of the configuration options to their inner models.
	if "EncoderDecoder" in self.__class__.__name__:
	config = None
	else:
	config = self.config

However, in some places, our codebase still try to access some attributes at the top level of the configuration (i.e. not inside the 2 components), like

transformers/src/transformers/models/vision_encoder_decoder/modeling_tf_vision_encoder_decoder.py

Lines 664 to 669 in 95e7057

	def serving_output(self, output):
	pkv = tf.tuple(output.past_key_values)[1] if self.config.use_cache else None
	dec_hs = tf.convert_to_tensor(output.decoder_hidden_states) if self.config.output_hidden_states else None
	dec_attns = tf.convert_to_tensor(output.decoder_attentions) if self.config.output_attentions else None
	enc_hs = tf.convert_to_tensor(output.encoder_hidden_states) if self.config.output_hidden_states else None
	enc_attns = tf.convert_to_tensor(output.encoder_attentions) if self.config.output_attentions else None

In particular, self.config may not have use_cache, for example, for this checkpoint "nlpconnect/vit-gpt2-image-captioning". We should instead look self.config.deocder.use_cache.

This PR try to follow the rule of # Encoder Decoder models delegate the application of the configuration options to their inner models..

This PR is also (another) one necessary step to fix #22731.

[ydshieh]

Use the attributes in the components' configurations (i.e. self.config.encoder and self.config.decoder)

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[sgugger]

Looks good to me but would like for a TF expert to have a look too!

[Rocketknight1]

This makes sense to me!