Subgradient methods ralgb5 and ralgb4 for minimization of ravine-like convex functions. We consider properties of the three computational forms of the r-algorithm proposed by N.Z. Shor for optimization of non-smooth functions that differ in the complexity of a single iteration. Discussed is a variant of the r-algorithm with adaptive stepsize control along the direction of the normalized antisubgradient in the transformed space of variables. The Octave functions ralgb5 and ralgb4 are described, which implement two computationally stable forms of the r-algorithms with adaptive stepsize control and a constant space dilation factor. The results of computational experiments for an essentially ravine-like piecewise quadratic function and a piecewise linear function related to solvability of interval linear tolerance problem are presented