204b4cad88ff526006e2fa2c6604c35902cf5aef,tensorflow_transform/analyzers.py,_QuantilesCombinerSpec,extract_output,#_QuantilesCombinerSpec#Any#,774

Before Change

    // All relevant state about the input is captured by "summary"
    // (see comment in add_input() and merge_accumulators()).
    with self._session.graph.as_default():
      self._session.run(
          self._qaccumulator.add_prebuilt_summary(
              stamp_token=self._stamp_token, summary=tf.constant(summary)))
      self._session.run(
          self._qaccumulator.flush(
              stamp_token=self._stamp_token,
              next_stamp_token=self._stamp_token))
      are_ready_flush, buckets = (
          self._qaccumulator.get_buckets(stamp_token=self._stamp_token))
      buckets, _ = self._session.run([buckets, are_ready_flush])

    // Quantile boundaries is a list of the form
    //    [np.ndarrary(min, <internal-boundaries>, max)]
    // The approximate quantile library can return less or more than requested
    // number of buckets. The max value can be same as the last internal
    // boundary, due to removal of duplicates.
    // Below, the min and/or max quantile boundaries are trimmed depending
    // on the actual boundaries returned by the library.
    if buckets.size >= (self._num_quantiles + 1):
      // Trim min/max.
      buckets = buckets[1:-1]
    elif buckets.size == self._num_quantiles:

After Change

    // (see comment in add_input() and merge_accumulators()).
    self._session.run(
        self._add_prebuilt_summary_op,
        feed_dict={self._prebuilt_summary_input: summary})
    self._session.run(self._flush_op)
    buckets = self._session.run(self._buckets_op)

Instances