LogText = regexp( fileread('YourLogFile.txt');, '\n', 'split');
NumericFields = regexp(LogText, '-?\d[^-+.e0-9]*', 'match');
NumericFields will then be a cell array with one entry per line of the original file (and possibly one extra entry at end of file). Each cell array entry will then be a cell array of strings, one numeric string per entry. So NumericFields{K} will be a cell array of strings, one string per numeric entry on the line.
Note that this has not been made selective about which lines are converted (as you did not give any indication as to which you wanted), so there will be some NumericFields{K} with one entry (e.g., Time line}, some with two (e.g., ExecutionTime line), a number with three per line with the source varieties not distinguished...
