Eric Meissner

Results 11 comments of Eric Meissner

Looks cool Tom, haven't had a chance to actually go through what the results look like yet but the changes to the core MXFusion codebase look fine to me.

I think this was happening before, I remember seeing it. The reason is that you're using Python 3.6 only string formatting in places. this "(f"Context device id {ctx.device_id} outside range...

Thanks for filing this! Not sure exactly where the fix is here, but after a little digging I would guess that the kernel constants parameters just aren't making it into...

Thanks, this definitely needs some improvement. It should be possibly to add something along the lines of ``` print(infr.params) > Variable(1ab23)(name=y) - (Model/Posterior(123ge2)) - (first mxnet values/shape) > .... ```...

Thanks for finding this! The first question I have for this is whether it even makes sense to pass PARAMETER type variables to the 'constants' argument in Inference? It may...

mxfusion/inference/inference.py:42 in the "discover_shape_constants" function.

We have a doc that explains this and a unit test that does it, we just need to put these together into a nicer notebook for people to find and...

Hi @gowtham1997, glad to see you're interested! The main place for this is the [Distribution class](https://github.com/amzn/MXFusion/blob/master/mxfusion/components/distributions/distribution.py#L105). Each Distribution subclass (e.g. [Normal](https://github.com/amzn/MXFusion/blob/master/mxfusion/components/distributions/normal.py)) will either be reparameterizable or not. This is based...

This is a bug on that line of code. It should probably look something: ```python new_leaf = v.replicate(var_map=var_map, replication_function=lambda x: ('recursive', 'recursive')) new_leaf.graph = new_model.components_graph ```

Looks fine to me. With another call to InferenceParameters inside that, similar to [fix_all](https://github.com/amzn/MXFusion/blob/develop/mxfusion/inference/inference_parameters.py#L139) but only fixing a single parameter. One question I have is around the difference between model...