django

 attention span is the real issue i think, 

everyone i know or see are doing something, why there aren't anyone that does nothing actually nothing for short long time like a hour. mobile, scroll, fidgeting, kei na kei garxa garxa why/??

i don't know how robot will turn out when i integrate these features but this thing is driving me very much, I am genuinely think that this can be amazing ufckin

integrate it tommorrow (one after another feature integration approach)

sdc (multiple camera stopping issue when used the display fuck it how to  solve find it.) 

robot ma feedback loops and reinforcement learning is most important aspect for the type of brain we want. 

reinforcment learning with human feedback isn't that feasible i think requires a bit dig (1 hour). 

if its feasible then surely the wait won't be too long when a working prototype is developed for further polishing and beginning other aspects of this robot. 

i don't know what this drive is, i think this is what i should be putting my brain mind everything into, let's do it where does it takes, more people, i will get the helping hand for sure. 

flash attention:  is a technique that makes transformer models faster and more memory-efficient by minimizing slow data movement between memory and processing units. It does this by loading all the query, key, and value data once into fast on-chip GPU memory, breaking it into smaller blocks for efficient processing, and fusing multiple operations into a single step.

vector database rey? 

similarity between the vectors in high dimensional space

check this website (aialignmentforum.org)

https://www.alignmentforum.org/library

started learning and implementing robotics, ros v-slam, kumari to main block (sdc)

tested orb_slam3 package inside ros, have some issue with the camera driver as the camera doesn't support the mono8 or rgb8 format with opencv, need to solve this issue.