As you said, the equation is separable:
d<em>y</em>/d<em>t</em> = 2<em>t</em> / exp(<em>y</em>)
exp(<em>y</em>) d<em>y</em> = 2<em>t</em> d<em>t</em>
Integrate both sides:
∫ exp(<em>y</em>) d<em>y</em> = ∫ 2<em>t</em> d<em>t</em>
exp(<em>y</em>) + <em>C</em>₁ = <em>t</em> ² + <em>C</em>₂
Move the constant terms to one side. When you add them together, you get another constant, so you can ignore the subscript altogether:
exp(<em>y</em>) = <em>t</em> ² + <em>C</em>
Solve for <em>y</em> explicitly by taking the logarithm of both sides:
ln(exp(<em>y</em>)) = ln(<em>t</em> ² + <em>C </em>)
<em>y</em> = ln(<em>t</em> ² + <em>C </em>)
<em>C</em> can be any number; if it happens to be 0, then you have
<em>y</em> = ln(<em>t</em> ²) = 2 ln(<em>t</em> )
so B is the correct choice.
You can also approach this from the opposite angle: Assume <em>y</em> is one of the given solutions, then substitute it into the ODE. (Bit more trial-and-error involved, so not a good idea if you're short on time.)
For example, if <em>y</em> = 2 exp(<em>t</em> ) as in choice A, you have d<em>y</em>/d<em>t</em> = 2 exp(<em>t</em> ), so the ODE would become
2 exp(<em>t</em> ) = 2<em>t</em> / exp(2 exp(<em>t</em> ))
which is clearly (I hope) not true.