Unity ML-Agents - Watch this first! (Is this framework for you?)

[@JadenWilliamsLIVE]

Very insightful. 👍

[@keyhaven8151]

I have always had a question about mlagents: they randomly select actions at the beginning of training. Can we incorporate human intervention into the training process of mlagents to make them train faster? Is there a corresponding method in mlagents? Looking forward to your answer.

[@neuro_corgi]

Actually, ML-Agents can be used for imitation learning, and imitation learning is a supervised learning

[@TylerGreen]

Really nice stuff! I plan on using ML for a future vid (probably in 2 vids) and have been watching your vids. They're super helpful!

[@ProjectWander]

This is amazing. I just starting building some trees in behavior designer, but without a TON of code its always going to look super robotic and end up doing the same predictable stuff every time. This looks like quite the solution.

[@Rizzmaster9001]

how did it work out for you?

[@ProjectWander]

@@Rizzmaster9001 I ended the project, but it could have worked. It takes a LOT of training

[@choco2482]

is it possible to export my ML model into other languages?

[@lucutes2936]

any alternatives?

[@anonymosranger4759]

Turns out he is a PyTorch Guy, Where is the Tensorflow Squad? (I am Tensorflow!)

[@juleswombat5309]

But doesn't ML Agents use Tensorflow anyway, I presume all the use of Tensorboard, inferred a Tensorflow backend. But the stack is pretty much hidden by the ML Agents hyperparamters and config files anyway. As a novice in Deep ML, I started with Keras, and hence now Tensorflow 2.0. But I am seeing a lot of the new Deep RL algorithms examples being developed and explained through PyTorch based networks, with easier control over which sub networks are frozen. I perceive that Academia has some preference for PyTorch.

[@lucutes2936]

Well you reached 10k subs

[@munyunu]

Thank you so much.. wow!

[@timothytyree5211]

Don't do it! Use numpy+matplotlib instead!