As someone who is currently waist-deep in post-graduate education here is my two cents:

None of these proposals matter, nor will any changes or improvements to them. The bottom line is that as long as there are too many scientists present the competition to stay alive will be intense and therefore conditions will remain poor. NIH/NSF proposals like this attempt to legislate behavior without recognizing the realities on the ground.

The lack of attention to graduate/post-graduate training is borne from two areas: 1) PIs are pushed by the system to extract every bit of effort possible effort from employees. The penalty for not doing so is a loss of funding and the end of a career. 2) As long as there are multiple applicants for every scientist/PI position there's less need to insure development of most individuals. As an example, one can simply ignore the bottom sixty percent of CVs to no ill effect when considering tenure-track positions. Those that remain will no doubt be pretty impressive. In this respect, post-graduate training is less an education than it is a 'selection' process in the biological sense. Put another way: why bother training postdocs, when most of them will fall out of the "system" anyway? The best will claw their way to success on their own and the rest are irrelevant. (This is a sentiment I've heard multiple times from faculty at my current Tier-I, top-20 research institute.)

Until the supply of PhD scientists comes back into line with demand working conditions will remain poor, salaries low, and hours long. In absence of a solution to the supply/demand imbalance all attempts to solve the resultant phenotypes will fail.