-
-
Notifications
You must be signed in to change notification settings - Fork 199
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[WIP] API for BNNODE #724
[WIP] API for BNNODE #724
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
AI-Maintainer Review for PR - [WIP] API for BNNODE
Title and Description ⚠️
Scope of Changes ❓
Documentation ⚠️
struct BNNODE <: NeuralPDEAlgorithm
struct BNNPDE <: NeuralPDEAlgorithm
function DiffEqBase.__solve(prob::DiffEqBase.AbstractODEProblem, alg::BNNODE,args...; )
These entities should have docstrings added to describe their behavior, arguments, and return values.
Testing ⚠️
Suggested Changes
- Please provide a more detailed description of the changes and how they address issue API for BNNODE. #719.
- If the changes related to the BNNPDE class and related functions are addressing a different issue, please consider separating them into a different pull request.
- Add docstrings to the
BNNODE
andBNNPDE
structs and theDiffEqBase.__solve
function to describe their behavior, arguments, and return values. - Include information about how you tested the changes, including specific test cases used, the testing environment, and any relevant test results or observations.
Reviewed with AI Maintainer
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
AI-Maintainer Review for PR - [WIP] API for BNNODE
Title and Description ⚠️
Scope of Changes 👍
Testing ⚠️
Documentation ⚠️
struct BNNODE <: NeuralPDEAlgorithm
struct BNNPDE <: NeuralPDEAlgorithm
function DiffEqBase.__solve(prob::DiffEqBase.AbstractODEProblem, alg::BNNODE,args...; )
Please add docstrings to these entities to describe their behavior, arguments, and return values.
Code Changes ⚠️
Potential Issues ⚠️
Reviewed with AI Maintainer
Looking good, I wanna see your tests passing and documentation with an example then lgtm |
should work
@Vaibhavdixit02 @xtalax on my local machine all tests pass(v1.9.3) but for the v1 and v1.6 marginal errors crop up, what should I do? Using earlier versions affects performance. Also the v1.6 logging CI tests gives the statistics compat error as the AHMC version I'm using(NeuralPDE allows v0.5) works only with the latest Julia versions(refer https://julialang.slack.com/archives/CN04R7WKE/p1693333431525009?thread_ts=1693333431.525009&cid=CN04R7WKE). |
Maybe we can drop 1.6 for this test group (it wouldn't be straightforward though). I don't recall why the compat was needed, is it possible to get rid of that? |
Looking at test failures the values seem to be in the right ballpark so my guess is this is just a numerical difference between your local runs and the CI machine. You are probably not on the same OS (linux)? Relax the tolerances of the tests |
was done here by @ChrisRackauckas . |
whoa! after this update in Build to the latest Julia version the tests are almost exactly how i wanted them to be! |
@Vaibhavdixit02 ive added the "missing" Likelihood term i was talking about, if you want to run it with the regular tests just uncomment line 53 in the file "advancedHMC_MCMC.jl", ive tried, it works! |
…ce for special likelihood term
Need to update docs for diff training strategies,etc
@Vaibhavdixit02 lgtm? |
|
||
function BNNODE(chain, Kernel = HMC; strategy = nothing, | ||
draw_samples = 2000, | ||
priorsNNw = (0.0, 2.0), param = [nothing], l2std = [0.05], |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why is this [nothing]
and not just nothing
?
phystd = [0.05], dataset = [nothing], | ||
init_params = nothing, | ||
physdt = 1 / 20.0, nchains = 1, | ||
autodiff = false, Integrator = Leapfrog, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All the NUTS kwargs should go into a struct or a namedtuple
A couple of minor things, but they can be addressed in a different PR since this one's quite big already. So LGTM |
Thank you 🎉 |
for issue #719