// the buffer, so that UnreadRune and UnreadByte can check for
// invalid usage. opReadRuneX constants are chosen such that
// converted to int they correspond to the rune size that was read.
type readOp int8

// Don't use iota for these, as the values need to correspond with the
// names and comments, which is easier to see when being explicit.
const (
	opRead      readOp = -1 // Any other read operation.

* `tf.lite`
    * Quantization for `FullyConnected` layer is switched from per-tensor to per-channel scales for dynamic range quantization use case (`float32` inputs / outputs and `int8` weights). The change enables new quantization schema globally in the converter and inference engine. The new behaviour can be disabled via experimental flag `converter._experimental_disable_per_channel_quantization_for_dense_layers = True`.