Update links

[schlichtanders]

The link in the blog-post https://fluxml.ai/2019/03/05/dp-vs-rl.html
as well as the links in this README.md

Note:
The code corresponding to this blog post can be found here

Are pointing towards old code https://github.com/FluxML/model-zoo/blob/10bd26ca21079b1c6ee246dd1938beab55178949/games/differentiable-programming/trebuchet/DiffRL.jl

That code is no longer working, and hence it should be pointed to the more current version https://github.com/FluxML/model-zoo/tree/master/contrib/games/differentiable-programming/trebuchet

[asuresh213]

Running the current version raises the following error.

ERROR: LoadError: Compiling Tuple{typeof(Trebuchet.shoot),Tuple{Float64,Float64,Float64}}: try/catch is not supported.

(Most likely rising from the computing the gradient inside of Flux.train!)

Flux.gradient((wind, ang, mass) -> Trebuchet.shoot(wind, ang, mass)[2], -2.0, 45.0, 200.0)

from the blog post also returns the same error.

ERROR: LoadError: Compiling Tuple{typeof(Trebuchet.shoot),Tuple{Float64,Float64,Float64}}: try/catch is not supported.

[Roboneet]

The current version raises the error because computing gradient inside a try/catch is not supported. It's best to keep track of that issue in model zoo as the code resides there

[schlichtanders]

the post from @asuresh213 was completely unrelated to my original question. @asuresh213 Please don't hijack other person's issues.

Anyway, @Roboneet can you update the links, at least in this repo's README respectively?
Of course it would also be great to update the link in the original blog post. Do you know where to address this?

[asuresh213]

???? @schlichtanders
Running the current version that you pointed to, raises try/catch errors. The code from the link you posted is not working. That's all I was trying to get across.

[schlichtanders]

@asuresh213 sorry for misunderstanding, you tried this code https://github.com/FluxML/model-zoo/tree/master/contrib/games/differentiable-programming/trebuchet, especially https://github.com/FluxML/model-zoo/blob/master/contrib/games/differentiable-programming/trebuchet/DiffRL.jl and it didn't work for you when instantiating the manifest.toml?

It worked for me.

Also the try/catch is taken care of in the use of forwarddiff, so I am confused

[Roboneet]

@schlichtanders The blog post is here. I forgot to mention, but the link in the README is updated.

[schlichtanders]

@Roboneet thanks a lot, I will open a similar ticket on the blogpost github then