18:51 <akhunti1[m]> Hi All ,

18:51 <akhunti1[m]> if you see line number 72 , classify method i am passing json_str as a test data for inferencing , instead of passing testdataset .

18:52 <akhunti1[m]> because i need to pass test data from curl api like this : curl --http0.9 -X POST -H 'Content-Type: application/json' -d'{"data": { "ndarray": [[1,2,3,4,5,6]]}}' http://localhost:9000/api/v1.0/predictions for inferencing .

18:53 <akhunti1[m]> but it is throwing me error .

18:53 <akhunti1[m]> could you please suggest me what changes should i do , for this .

19:27 <akhunti1[m]> This is the error message i got . As my requirement is i need to pass from curl command for inferencing like this [ curl --http0.9 -X POST -H 'Content-Type: application/json' -d'{"data": { "ndarray": [[1,2,3,4,5,6]]}}' http://localhost:9000/api/v1.0/predictions ] ,

19:28 <akhunti1[m]> so i cannot hard coard test data set like line number 64 .

19:31 <rcurtin[m]> so the data comes as a string? like the string [[1, 2, 3, 4, 5, 6]] represents the data?

19:32 <akhunti1[m]> yes rcurtin

19:34 <rcurtin[m]> ahh, you will need to parse it then. it could be a little tedious, but I suggest a two-pass approach, where in the first pass you compute the size of the matrix, and then you create an arma::mat of the required size, and then in the second pass you fill it

19:35 <rcurtin[m]> you could consider using C++'s istringstream to parse the input, or maybe strtok(). I don't have any code samples handy, it could be a little tedious to do, but that is what you'll have to do

19:35 <rcurtin[m]> or... wait a second, I know that Armadillo has a string constructor. I am not sure if it will work exactly but you could try arma::mat(data) to see if it works

19:35 <rcurtin[m]> https://arma.sourceforge.net/docs.html#Mat

19:36 <rcurtin[m]> check out the section on the mat(string) constructor. you will probably have to do some reformatting of the string that you get as input, but that would likely be a lot easier

19:48 <akhunti1[m]> Hi rcurtin Thanks [ arma::mat(data) ] this line solved the problem .

19:53 <rcurtin[m]> make sure to double check that it is actually parsing the data correctly; it would be easy for that constructor to actually fail and, e.g., produce an empty matrix 👍️

19:59 <vaibhavp[m]> rcurtin: Hi! If you have time can you please look at the demo for DAGNetwork API and share your thoughts on it? Code for DAGNetwork class is the same repo. It's still a little rough but gives the overview. I would be very grateful. [ResNet_Example](https://github.com/mrdaybird/mlpack-dag-network/blob/main/ResNet_Example.cpp)

20:23 <akhunti1[m]> Hi rcurtin

20:23 <akhunti1[m]> i got the out put like this

20:27 <vaibhavp[m]> akhunti1: You can also try to print the data matrix(from arma::mat(string) to check that if it is the correct matrix or not.

20:30 <akhunti1[m]> sure , Thanks vaibhavp

22:46 <DavidValenteGSoC> Hello, my name id David Valente, I´m in my 3rd year of CSE at IST Lisbon and I´m interested in contributing to the algorithm optimization project for GSoC. Are there any recommended issues I could help to get started with? Thanks