Oh. So say your file (made of your vocab tokens) is some code between 1-243. Your vocab tokens, some are more frequent in all past observations, ex. 'the' not 'boat'. So common tokens get shorter codes basically. So your 1-243 code can be shorter/longer based on this. The range you go would be ex. 1-200 if a token is your most frequent token in vocab, so the code doesn't need to squeeze in little ranges like 0.85182572638756, it can just go 0.2504300. I'm not sure how better this is than the optimal example above yesterday, because huffman variable codes would be horrible ex. the=0, he=10, we=110, hi=11100, she=11101, joy=11110, mars=11111000.....these 8 bits could give us 256 codes.
As for the OP's question, Python, C++ etc are all languages that let you program your computer, which is simply a Turing Tape really. So the possible algorithms, although look like a spaghetti (change one thing and it does something totally different), are all just doing the same thing! Like a Turing Tape they all decide what bit they are on reading, what state they are in, and then what bit to write, what state to go to, and where to look next (left or right). So to permutate it along the line if you get me one by one, you would want to consider the state it's in, and the bit it is on, because these decide what to write, what state to go to, and where to look! But this won't do. What are their meanings? The human brain is where a algorithm is designed, so a human brain is a complete Turing Tape and can do anything the computer Turing Tape can do. So we should describe all algorithms using a human brain instead. So we need a AGI, architecture. We designed words to mean cat, horse, animals, they are quantinized and have Ranges. So say you want a similar algorithm to your algorithm, look in the brain to get it! ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T56d8c6758403c8e0-Mbfc5b428956474d28fbb795b Delivery options: https://agi.topicbox.com/groups/agi/subscription
