divergence on gluon-gluon to WW -> lvjj

Dear MG authors,

I am trying to g g > w+ w- > l v~ j j process via loop-induced mode. Here are links to my run/proc cards[1]

1)The problem is that if one of W is set to decay to j j, the amplitude diverges, - the syntax I used is

import model loop_sm-ckm_no_b_mass

define ell+ = e+ mu+ ta+
define ell- = e- mu- ta-

generate g g > w+ w-, w+ w- > ell+ vl j j $$ h [noborn=QCD] @0
add process g g > w- w+, w- w+ > ell- vl~ j j $$ h [noborn=QCD] @1

2)To check the divergence is due to decay of W, I remained W not to deacy.
generate g g > w+ w- $$ h [noborn=QCD] @0
add process g g > w- w+ $$ h [noborn=QCD] @1

Then, I can get stable result. I guess the divergence is caused by decay of W boson.

3)To check the diverge is by low mass offshell w, I set bwcutoff = 9999999. I can check the result is the same with 2).

4).Currently, I set min pt of lepton and jet to 10 GeV both.(you can find this setup in my runcard)
But the divergence is still remained.
MG shows the event record with divergence like below:

WARNING: "One Event with large weight have been found (ratio = 2.01e+05) in channel G1.22 (with rel.contrib=0.902).

<event>
 9 1 +1.8291740e+04 1.52999000e+02 7.54677100e-03 1.09458200e-01
21 -1 0 0 503 502 +0.0000000000e+00 +0.0000000000e+00 +9.5453957388e+00 9.5453957388e+00 0.0000000000e+00 0.0000e+00 -1.0000e+00
21 -1 0 0 502 503 -0.0000000000e+00 -0.0000000000e+00 -6.1308849026e+02 6.1308849026e+02 0.0000000000e+00 0.0000e+00 1.0000e+00
23 2 1 2 0 0 +8.8817841970e-16 -1.6653345369e-15 -6.0354309453e+02 6.2263388600e+02 1.5299898382e+02 0.0000e+00 0.0000e+00
-24 2 3 3 0 0 +6.5665704185e-03 +3.2687045683e-03 -3.0050481513e+02 3.1080480010e+02 7.9350361030e+01 0.0000e+00 0.0000e+00
24 2 3 3 0 0 -6.5665704185e-03 -3.2687045683e-03 -3.0303827939e+02 3.1182908590e+02 7.3519929147e+01 0.0000e+00 0.0000e+00
15 1 4 4 0 0 +5.0148557003e+00 +3.2523860278e+01 -2.3701410278e+02 2.3929436456e+02 1.7770000000e+00 0.0000e+00 -1.0000e+00
-16 1 4 4 0 0 -5.0082891299e+00 -3.2520591574e+01 -6.3490712346e+01 7.1510435536e+01 0.0000000000e+00 0.0000e+00 1.0000e+00
4 1 5 5 501 0 +3.0338863630e+01 +5.9783416941e-01 -2.3948949783e+02 2.4140427424e+02 0.0000000000e+00 0.0000e+00 -1.0000e+00
-3 1 5 5 0 501 -3.0345430201e+01 -6.0110287398e-01 -6.3548781565e+01 7.0424811659e+01 0.0000000000e+00 0.0000e+00 1.0000e+00
<mgrwt>
<rscale> 2 0.15299898E+03</rscale>
<asrwt>0</asrwt>
<pdfrwt beam="1"> 1 21 0.14685220E-02 0.15299898E+03</pdfrwt>
<pdfrwt beam="2"> 1 21 0.94321335E-01 0.15299898E+03</pdfrwt>
<totfact> 0.17112983E+06</totfact>
</mgrwt>
</event>

Link is also here[http://147.47.242.40/MG/ggWW/cards/div_event.txt]
It seems like
 g g > quark loop > Z -> WW -> lvqq

with Z boson highly boosted along z direction and its daughters(W-,W+) have very small angle to each other.

Could you point which part of run card has problem and causes this divergence?

If you have anything unclear, please let me know. Thank you so much!

Best,
Junho

[1]http://147.47.242.40/MG/ggWW/cards/ggWW_run_card.dat
[2]http://147.47.242.40/MG/ggWW/cards/ggWW_proc_card.dat