The assessment is as follows: To be corrected.

The paper is well written and can be published in the SQM proceedings.

The authors should address the points below to improve the manuscript.
(Some are mere typos, since I understand there will be no copy editing
 going forward.)

A. Thank you very much for the quick and careful check.
Please find our answers below embedded in your comments.

1) The two bands in Fig. 1 (CGC-NRQCD and NLO NRQCD) are
indistinguishable in a black and white print out. Would be nice if they
were made a little more different. Also referring to colors in the caption
is not the best practice for the same reasons.

A. We have updated the Fig.1 to make it distinguishable in a black and
white print out. The solid band of NLO NRQCD is changed to the hatched band.
We also updated the legends in Fig. 4, 5, and 6 to more explicitly
acknowledge theory groups. Regarding referring to colors in figure captions,
we also mention the marker style in addition to marker color, which should
avoid confusions in this proceedings.

2) What is B on the y axes in Figs. 1 and 2 (B_|| in Fig. 3)?

A. We now refer to the branching ratios B and B_{ll} in the figure
captions 1, 2, and 3.

3) Discussion of Fig. 4: Transport model II seems to describe R_AA even
quantitatively.
A. Please find the answer to the comment 5) below.

4) Discussion of Fig. 5: "...but tends to overestimate ..." -> "...but
tend to overestimate
A. Fixed.

5) Fig. 6: Model I seems to do fine, no? Text mentions "tension" there. (again in Summary)
A. We would like to answer to your comments 3) and 5) together here.
We prefer to keep the statement as it is. Despite that in Figure.4 TAMU
calculation (Model II in the ver4 pdf) can qualitatively describe RAA as
a function of pt, this is not the case  if we select more finer
centrality bins as shown in back-up slides of  the conference talk.
Also in Figure 6, Tsinghua calculation (Model I in the ver4 pdf) can
quantitatively describe peripheral collision data, while it over predicts
data about a 1.5 sigma at Npart~240. Conclusions from model comparisons 
to data at high pt, therefore, can be quite sensitive to  selections of
centrality and pt cut off etc.. Having said that, in this proceedings,
we would like to avoid making strong comments about which model calculation
works better. However, it is certainly true that our new data provide existing
models with more constraints.