ml5-data-and-models

ml5-data-and-models copied to clipboard

default branch changing to main

1

I am doing a little bit of cleanup on this repo as related to my [Intro to ML course](https://github.com/ml5js/Intro-ML-Arts-IMA-F20), specifically around example pre-trained `charRNN` models. I have updated the default...

shiffman

redid the yolo model

1

* removed batch norm layers as they are not needed for inference ( batch folding) [here](https://github.com/ModelDepot/tfjs-yolo-tiny/issues/6#issuecomment-388422065) * rewrote the yolo model.json file , it's now more human friendly (removed some...

hiddentn

Image License

1

Hi, I want to use these few of the landscape images for just for some entropy calculation (no modification, no sharing etc). What is the licensing procedure for it? and...

shwetasaharan

Add a subset of Google Quickdraw data?

1

This relates to what I'm working on in terms of custom Image Classification: https://github.com/ml5js/ml5-library/issues/121. I think we can link out to MNIST, but not necessarily feature it here. (I'm writing...

shiffman

LSTM train.py not loading from checkpoint

3

Hi, I try to start training a model from a checkpoint. I use `init_from` option but somehow the model still starts training from 0. What could be the problem here?...

karooolis

Using `ml5.LSTMGenerator` with the [shakespeare](https://github.com/ml5js/ml5-data-and-training/tree/98de6599693e870afb0d53198b2585e3da5aed0b/models/lstm/shakespeare) model generates text in which everywhere a lowercase letter `u` should appear, two space characters appear instead. Uppercase letters `U` are missing completely. This is...

cwalther

Google Colab as platform for cloud training of models

4

This comes from [ml5.js#55](https://github.com/ml5js/ml5-library/issues/55): (Yes, eventually we want to train models in JavaScript in the browser too, but this might be useful now and later.) https://research.google.com/colaboratory/unregistered.html

cvalenzuela

Right now we link out to the blog post about the landscapes dataset. I'm wondering if we should excerpt some of the tutorial / explanation content here as part of...

shiffman

Piggybacking off of #9, it might be nice to include a variety of pretrained word vectors. Also should double-check the licensing / credits on the current provided word vectors and...

shiffman