so far what i have understand from the gym api(car game) docs that env.step() returns reward either -1(until car position<0.6) or 0(once it is >=0.6 means completed) mujhe ye click nahi ho rha like at a particular state how can i say yess the taken action is worth or not.
like reward will always be -1 until whole completion and in this situtation how can i take reward as an output for the given state, as it will always be -1.
how do i train my model?