DataPredict [Release 1.21] - General Purpose Machine Learning And Deep Learning Library (Learning AIs, Generative AIs, and more!)

Resources Community Resources
, ,
#723

Also, please don’t use strings for your class label. Instead use 1 and -1.

#724

i did that in the debugging part

#725

DataPredict Neural Beta Release!

Hi guys! I am excited to announce that I have released a PyTorch-like Library for Roblox and Pure Lua.

You can have a look here!

5 Likes
#726

Hey guys, I want to let you guys know that I made minor code refactoring to fix some issues for Release 1.16 version of the library. I recommend you to update the library as soon as possible.

3 Likes
#727

Hey guys! I renamed some of the functions in the MatrixL for less verbosity and required some refactoring in the DataPredict library.

It’s not urgent to update the library, but if you find yourself getting errors before updating any libraries, then all you need is to update both of the libraries at the same time.

1 Like
#730

Release Version 1.17

Changes

  • Refactored RandomNetworkDistillation so that it no longer inherits NeuralNetwork. Please use setModel() function to add neural network model to it. Also added deep copy option for TargetModelParameters and PredictorModelParameters.

  • Regularization no longer need number of data, instead the gradient method models will handle the operations related to it instead.

  • Improved calculations related to calculation of cost function derivatives involving both regularization and optimizers.

Fixes

  • Made some bug fixing.

Removed

  • LongShortTermMemory and RecurrentNeuralNetwork models are removed as these do not work well with the current API.

Note

These changes are for preparing long-term maintenance mode and will no longer receive new updates or features. In other words, this library will soon reach End-Of-Support.

If you want a more advanced deep learning library, you can have a look at DataPredict Neural here.

In the future, LSTM and RNN will be added to the DataPredict Neural library to replace the models that are being removed from the current DataPredict library.

3 Likes
#734

Hello guys! I made some fixes to affinity propagation model and made some changes.

If anyone is using that model, I highly recommend you to replace the current version of Release 1.17 to updated version of Release 1.17 as soon as possible.

1 Like
#735

Hey so i just downloaded version 6 because i was very intrigued. However i seem to get this error when starting:

image
image1444×372 76.6 KB

I know im probably missing something so i came to you.

#736

Oh sorry about that! I was just bug hunting and somehow the internal test tools didn’t catch this.

I have pushed the fix to GitHub.

#737

Hello guys! I made some fixes and improvements to all experience replays.

If anyone using any kind of experience replay, I highly recommend you to replace the current version of Release 1.17 to updated version of Release 1.17 as soon as possible.

2 Likes
#738

Hello guys! I made some improvements to the neural networks!

I recommend you guys updating Release 1.17 as soon as possible. Also, this should make the sword-fighting AIs train slightly more faster.

1 Like
#739

Hello, thank you so much for making this! This is incredible, and i will definetly use this in my projects.

2 Likes
#741

im making a new Anti-Cheat now, but why does the cost value just go up? ive tried different combos of different MatrixL and DataPredict versions but nothing works, heres the code:

local module = {}
local MT = {}
MT.__index = MT

function module.new(Dataset, Varibles, Externals, Mask)
	local self = {}
	setmetatable(self, MT)
	local NewDataset = {}
	local Labels = {}
	for i, v in pairs(Dataset) do
		local Div = 0
		for i2, v2 in pairs(v["Divs"]) do
			Div += v2
		end
		Div = Div * v["Time"]
		local Statistics = {}
		for i2, Stat in pairs(v["Stats"]) do
			table.insert(Statistics, Stat/Div)
		end
		for i2, v2 in pairs(Externals) do
			if v["Externals"][v2] then
				table.insert(Statistics, v["Externals"][v2])
			else
				table.insert(Statistics, Mask)
			end
		end
		if v["Cheater"] == true then
			table.insert(Labels, {1})
		else
			table.insert(Labels, {-1})
		end
		table.insert(NewDataset, Statistics)
	end
	print(NewDataset)
	local Model = require(game.ReplicatedStorage.DataPredict.Models.SupportVectorMachine).new(100, 1, 'RadialBasisFunction')
	Model:train(NewDataset, Labels)
	return self
end

return module
#742

Hello guys! I added two more models which are double dueling Q-learning variants. I also made some improvements to the library.

I recommend you guys updating Release 1.17 as soon as possible.

1 Like
#744

Good news people! You guys can now do accelerated self-learning AI training!

No more long training times!

3 Likes
#745

Release Version 1.18

Added

  • Added ValueScheduler class. These will help you adjust the values as you call the calculate() function.

  • Added setEpsilonValueScheduler() and getEpsilonValueScheduler() function into the ReinforcementLearningQuickSetup.

  • Added setLearningRateValueScheduler() and getLearningRateValueScheduler() function into BaseOptimizer.

Changes

  • Renamed setPrintReinforcementOutput() to setPrintOutput() for ReinforcementLearningQuickSetup.

Removed

  • Removed epsilon decay factor parameter inside the ReinforcementLearningQuickSetup in favour of using ValueScheduler.

  • Removed timeStepToDecay parameter from the LearningRateTimeDecay optimizer.

1 Like
#750

Hi. Does this allow reward learning?

#751

Somewhat. Currently the library only have Random Network Distillation if you want to do internal rewards.

#752

Im trying to create something like in those videos of how it has to walk if yk what i mean

#753

Possible, but quite limited with this library. Currently, it doesn’t support continuous action spaces. Only discrete ones.

Likely it won’t be implemented on this library, but rather on “DataPredict Neural” library.