You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi :)
Unfortunately I am not sure what you are asking - the number of parameters in a CNN is always dependent on the input size?
There are lots of examples if you google for how to compute the number of parameters (as for example https://towardsdatascience.com/understanding-and-calculating-the-number-of-parameters-in-convolution-neural-networks-cnns-fc88790d530d).
For the space those parameters then need on the AIdeck's memory it is important in which format they are saved - usually you'd want to quantize it to 8bits, so you'll need 1 byte per parameter. In L2 you of course also need space for the current and next activations.
Hope this helps,
Hanna
This discussion was converted from issue #582 on February 23, 2023 13:33.
Heading
Bold
Italic
Quote
Code
Link
Numbered list
Unordered list
Task list
Attach files
Mention
Reference
Menu
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Hi, I am quite new embedded systems which have limited memory.
I want to calculate the number of parameters of my custom network for deploying it to ai deck.
but it seems that the ai deck's memory is depends on the input size as well.
Can I get some guidance for that?
Thanks :)
Beta Was this translation helpful? Give feedback.
All reactions