Personally, i recommend reviewing some mathworks first before moving onto machine learning and furthermore deep learning, e.g., calculus, linear algebra and probability and statistics. And some basic elements of deep learning (maybe optimization strategy like SGD and Adam, loss functions etc.) These are the fundamentals, literally I got into trouble with stochastic gradient descent as first, not even mentioning backpropagation for updating the weights of FFN
1 Like
I think i need to understand both sorry for not being clear
Thank you so much, I’ll make sure to learn that
Since that other guy pretty much covered the theoretical aspect, I don’t really need to explain all that.
However, I would like to know if you have read the documentation already? Those contains important tutorials for you to cover. For beginners like you, all you need to cover the ones that are clearly marked “Level 1”, as well as the Neural Networks part.
Hello guys! I would like to inform you on the change of what we determine commercial use as to support smaller developers who only started off their game development business.
Here are the list of changes:
-
1000 USD in a life time to 1000 USD per year.
-
100 active players in a single time frame to no limit.
-
When a company has subsidiaries or related entities, the total revenue is combined.
Hopefully, by making the threshold more lenient, it would help their businesses to grow further and better in shorter amount of time.
how do i resolve this issue?
the npc models are spinning around randomly after few min it training, i had to restart to fix it.
It depends on what algorithm you use. I need more information than this to answer your questions.
Are you using the sword-fighting self-learning AIs code that was in my post?
yes im using the newest one v8 and also i tried to remove jump class action.
If the spinning around was as a result of removal of jump class action, I recommend you to remove the pre-trained model parameters from the models.
i did already and edited the addLayer from 7 to 6
I’m guessing you may need to modify the reward. It’s definitely learning reward hacking at this point.
so when the npc start walking around/spinning the model parameter saved is all nan and 0, i have to restart to fix it again and over, over, over…
Hold on, let me try to replicate this issue.
How long did you train it for before it starts spinning? Right now, I can’t replicate it. Also did you change the learning rate as well?
Also, I don’t generally recommend having layers that have the same number of neurons to the adjacent layers. Right now, you have 6 output and 6 hidden layer.
9 & 6 layer, still same 9 inputs and 6 output. i will give the file later.
Kk. I’ll have a look at that then.
hey, here…
training model 3.rbxl (340.1 KB)
just wait like few min until it start spinning and give the nan’s
Okay, it is replicated. The only strange thing is that when I modified my own file to follow yours, it didn’t run into this issue.
That being said, I think you should change the final layers for all model parameters to have LeakyReLU as their activation function.
Also, start off with new parameters that aren’t stored in datastores.
i already did and still the same thing.
then how much the layer will be to set.
Model:addLayer(3, true, 'LeakyReLU', 0.001)its this what you mean? but the includeRND is off by default and i also tried to change that to “4” and its now better but it still happens but just took longer than before…
and finally it seems back normally when set to 6 from 3/4
but why the highest reward is 0? is this correct? it happens multiple times in a row…
training after like 5-15 min. but after like 20-30 min later it start spinning again but just for short time like 4-7 sec and it back to normal again.
also this library is using native luau operation? if not maybe you can try, it might can gain some performance increases.
after restart the run, it spinning at first start…
im thinking its not the LeakyRLU (RND)
this is my latest file:
training model 3.rbxl (340.9 KB)
so i went redownload the file and only remove jump from the class and change the layer to 6 from 7, it seems to be fine but after few min the run start it start spinning again… and comparing this with previous one, im thinking the previous provide better result because they show the good performance by killing & movement to target, while this one is like just moving randomly and having high minus rewards.
once i tried to enable the includeRND variable, it immediately start spinning at start. i have no idea right now…
i dont know why each model save twicely when end the run.
training model 3.rbxl (340.9 KB)
when using rnd:
+the parameters is all nan