Well, I predict 256 letters, so, it is not 50/50 chance for me, it starts real bad, it is tough for it, eventually it learns the distribution though!
not using cmix preprocessor, I got 10MB 0.5505950784146294 1MB 0.5310066708037171 100KB 0.47092919877903217 10KB 0.46603868078103533 1KB 0.5588715348677878 Hmm, why are these scores better?, the preprocessor makes words into symbols ex. the into Y and walking into 8R......and I predict a letter at a time like that, 8...R....and so on..............oh i need to inflate maybe!........file size is not 100MBs after preprocessor, it's ~60MB, so .......... maybe if I predict walk by predict R=0.34% likely is next, I should times it as 4 samples then, 0.34, 0.34, 0.34, 0.34, for each letter in 'R', w a l k....this will require me to get into the code... so the detected issue here is I'm scoring using less samples...keep in mind I need score for each letter...sure predicted walk as R with 100% confidence is good but .....hmm.... ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Td13a829978c4c9f3-M0af34062c38d42e1e235d232 Delivery options: https://agi.topicbox.com/groups/agi/subscription
