How can I identify the number of records at each treatment level the teffects matching estimators use to estimate treatment effects?

Here is a nearest-neighbor example with two treatment levels illustrating the question.

Code:
. use http://www.stata-press.com/data/r13/cattaneo2, clear
(Excerpt from Cattaneo (2010) Journal of Econometrics 155: 138-154)

. tab mbsmoke

1 if mother |
     smoked |      Freq.     Percent        Cum.
------------+-----------------------------------
  nonsmoker |      3,778       81.39       81.39
     smoker |        864       18.61      100.00
------------+-----------------------------------
      Total |      4,642      100.00

. teffects nnmatch (bweight mage prenatal1 mmarried fbaby) (mbsmoke), nn(1)

Treatment-effects estimation                   Number of obs      =      4,642
Estimator      : nearest-neighbor matching     Matches: requested =          1
Outcome model  : matching                                     min =          1
Distance metric: Mahalanobis                                  max =        139
----------------------------------------------------------------------------------------
                       |              AI Robust
               bweight |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
ATE                    |
               mbsmoke |
(smoker vs nonsmoker)  |  -240.3306   28.43006    -8.45   0.000    -296.0525   -184.6087
----------------------------------------------------------------------------------------

.
.
. tebalance summarize
note: refitting the model using the generate() option

  Covariate balance summary
                                                   Raw      Matched
                          -----------------------------------------
                          Number of obs =        4,642        9,284
                          Treated obs   =          864        4,642
                          Control obs   =        3,778        4,642
                          -----------------------------------------

  -----------------------------------------------------------------
                  |Standardized differences          Variance ratio
                  |        Raw     Matched           Raw    Matched
  ----------------+------------------------------------------------
             mage |   -.300179   -.0040826      .8818025   .9815517
        prenatal1 |  -.3242695   -2.78e-16      1.496155          1
         mmarried |  -.5953009   -2.42e-16      1.335944          1
            fbaby |  -.1663271    2.24e-16      .9430944          1
  -----------------------------------------------------------------
In this example, mbsmoke is the binary treatment indicator and 864 of 4,642 records receive treatment, the remainder do not receive treatment. The nearest-neighbor estimator applies 1-1 matching with replacement and estimates an ATE of approximately -240. Eight hundred and sixty-four treated units (mbsmoke==1) and some number of matched control units (mbsmoke==0) are used to generate the ATE of approximately -240. The number of matched non-treated units could range from the single digits up to 864 because teffects matches with replacement.

Output from the teffects command shows that 4,642 is the "number of observations," which I believe means the number of records used by the matching procedure, not the number of records used to estimate the ATE. From what I can tell, the output from the "tebalance summarize" command does not tell me how many non-treated units are used to estimate the ATE.

How can I identify the number of records at each treatment level the teffects matching estimators use to estimate treatment effects?