lua - torch: what is the structure for tensors of different sizes? -
table? inserting big tensors table not efficient, not possible because of memory in case. works fine, it's ugly:
local s_ = 0 s_ = s_ + 1; local x_py_1 = fromfile(('%s/x_py_%.2f.bin'):format(data_dir, scales[s_])) s_ = s_ + 1; local x_py_2 = fromfile(('%s/x_py_%.2f.bin'):format(data_dir, scales[s_])) s_ = s_ + 1; local x_py_3 = fromfile(('%s/x_py_%.2f.bin'):format(data_dir, scales[s_])) s_ = s_ + 1; local x_py_4 = fromfile(('%s/x_py_%.2f.bin'):format(data_dir, scales[s_])) s_ = s_ + 1; local x_py_5 = fromfile(('%s/x_py_%.2f.bin'):format(data_dir, scales[s_])) s_ = s_ + 1; local x_py_6 = fromfile(('%s/x_py_%.2f.bin'):format(data_dir, scales[s_])) x_py = {x_py_1, x_py_2, x_py_3, x_py_4, x_py_5, x_py_6}
show code.. please you
x_py = {x_py_1, x_py_2, x_py_3, x_py_4, x_py_5, x_py_6} i, v in ipairs(x_py) v = fromfile(('%s/x_py_%.2f.bin'):format(data_dir, scales[i-1])) end
simply using table, want it?
Comments
Post a Comment