Commit History

making it a python package on PyPI
75481dd

Jensen-holm commited on

Moving towards making this a python package for simple neural network
03f2b37

Jensen-holm commited on

getting rid of regression tab, thinking about having a custom data tab
9117cbd

Jensen-holm commited on

fixing bug that happened with a type that I made, I think huggingface
de85694

Jensen-holm commited on

adding Sigmoid as an option for output layer activation
28a7ac6

Jensen-holm commited on

changing name to Numpy-Neuron
b471057

Jensen-holm commited on

getting rid of unessesary imports
e236c9a

Jensen-holm commited on

latest update, finishing classification with MNIST! More details on
6d0453c

Jensen-holm commited on

switching to gradio and a complete rewrite that I did over in the
6307b4f

Jensen-holm commited on

fixed bug that returned same number twice in response
d04aaf5

Jensen-holm commited on

also returning accuracy score from test data from the api
bff0f22

Jensen-holm commited on

ready for prod
336625e

Jensen-holm commited on

working on readme and adding ml vis submodule
3ca18fe

Jensen-holm commited on

adding accuracy score tracking
303df47

Jensen-holm commited on

adding CORS support
8ea5c07

Jensen-holm commited on

added loss history in the return to the user so they can make graphs and
2fc4a94

Jensen-holm commited on

must specify werkzwug for flask to work on render
ceca02c

Jensen-holm commited on

forgot to add gunicorn
843007e

Jensen-holm commited on

ignore and removing dockerfile
abcc2e0

Jensen-holm commited on

adding dockerfile
a4c4ba0

Jensen-holm commited on

adding accuracy score metric
03176c2

Jensen-holm commited on

moving on to testing other datasets!
9e506b7

Jensen-holm commited on

currently debugging mismatched shape error when computing weights and
880505a

Jensen-holm commited on

init weights and biases, and getting through epochs
84bbd7d

Jensen-holm commited on

init weights and biases
38e3b7b

Jensen-holm commited on

successful example
14841f9

Jensen-holm commited on

handling errors well with the neural netork api
8c348c5

Jensen-holm commited on

working on new python implementation with cleaner code
29cce3f

Jensen-holm commited on

initial dumb api working
2c781f8

Jensen-holm commited on

stuck on working with computeOutput function, getting dim error every
0987346

Jensen-holm commited on

mapping activation arg to the actual function
0a04cd7

Jensen-holm commited on

initializing weights and biases
59ae052

Jensen-holm commited on

working through Train() function
6b47160

Jensen-holm commited on

updated ignore for vim *.swp files when committing from vim terminal
c4864dc

Jensen-holm commited on

added trainTestSplit() function
554811e

Jensen-holm commited on

adding activation funcs
27c0b63

Jensen-holm commited on

updated example for new format
fae8f77

Jensen-holm commited on

significantly decreased the complexity of the API by deciding to make
eec3b88

Jensen-holm commited on

re configured the Alg interface and the NN sttuct, now we have to work
a4aa491

Jensen-holm commited on

splitting up code into seperate modules
dfd3544

Jensen-holm commited on

reads incoming request json if user specifys correctly
b5b4a38

Jensen-holm commited on

initial basic REST API setup
5d966be

Jensen-holm commited on

going to use this branch to re make this program in go using the gota
da3cde4

Jensen-holm commited on

switching back to bas64 encoding/decoding with image byte data
f308820

Jensen-holm commited on

re formatting the api to contain the image bytes in json
e571d8c

Jensen-holm commited on

changing the get plot route to return bytes of the image
d13c92f

Jensen-holm commited on

creating the plot endpoint for getting result images
f38f95f

Jensen-holm commited on