hi i am new to this community, but i came up with this curiosity as i am having an embedded board to train a sklearn i have nvidia jetson tx2 with 256 nvidia cuda cores

so inorder to take less load in ram i just transffered the training data into excel file and with its import i can train a model

importing training data

now inorder to give it to model there is one way by adding the data to list or numpy array or tensor

like wise follow up code excel data into data strucutre

and then giving it to the model by model.fit model.fit

i know this is one way which is not optimal

my 1st question is , is there any way exist to directly give an excel file to an .h5 model without any intermediate data structure to be neglected i tried to search online but didn't find any way if you find or know as a community we should discuss and contribute the knowledge. to learn and improve together.

now one secondary follow up question also exist in my mind which is

my 2nd question is how to write optimize code specially for the jetson devices or any embedded board , coming to time & space complexity as one of concern i write minimilstic code only also the nvidia compiler are already written to code optimisation but is there any way optimise even much code which i am lacking

my normal code looks like

just any minimlistic optimize code writing approach for a limited resource compute machine like embedded boards with minimal basic scratch library writiing like c++

  • $\begingroup$ Hello. Can you start your sentences with a capital letter, use I instead of i and also use punctuation? That would be great. Also, questions about specific libraries are off-topic here. $\endgroup$
    – nbro
    Commented May 18 at 13:54


You must log in to answer this question.

Browse other questions tagged .