Difference in precision of the probabilities & logits obtained

I followed the instructions & tried to run `$ python evaluate_sample.py`. But the results i obtained are slightly different . Please let me know how to proceed. These are my results:

Note: Highlighting a few of the discrepencies

Norm of logits: 138.**468613**

Top classes and probabilities
1.0 41.813667 playing cricket
1.4971563e-09 21.493969 hurling (sport)
3.8431092e-10 20.134098 catching or throwing baseball
1.549236e-10 19.225578 catching or throwing softball
1.13602086e-10 18.915348 hitting baseball
8.801189e-11 18.660118 playing tennis
2.4415699e-11 17.377872 playing kickball
1.15319716e-11 16.62777 playing squash or racquetball
6.1319474e-12 15.996159 shooting goal (soccer)
4.391761e-12 15.662376 hammer throw
2.2134141e-12 14.977182 golf putting
1.6307188e-12 14.671667 throwing discus
1.5456307e-12 14.618078 javelin throw
7.669135e-13 13.917265 pumping fist
5.192998e-13 13.527372 shot put
4.2681746e-13 13.331247 celebrating
2.**7205825**e-13 12.880907 applauding
1.8357085e-13 12.487492 throwing ball
1.6134758e-13 12.358452 dodgeball
1.1388438e-13 12.010075 tap dancing

Thank you!