Learning How Hard to Think: Input-Adaptive Allocation of LM Computation

Re-assign community