[FR] allow orthogonal parameters in pyro.param

Description

Contributor guide

Tech stack

pythonpytorch

Domain

machine learningbackend

Issue type

feature

DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.

3

Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.

half day

Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.

stale

ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.

clear

Prerequisites

familiarity with Pyro's param systembasic knowledge of PyTorch parametrizations

Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.

40

Research direction

Start by reviewing the implementation of `pyro.param` to understand how parameters are currently declared and constrained. Then explore `torch.nn.utils.parametrizations.orthogonal` to see how orthogonal parametrization is applied in PyTorch. Look for any existing discussion in the single comment on the issue for additional context on desired behavior. Consider implementing an optional `parametrization` argument in `pyro.param` that accepts orthogonal or other parametrizations from PyTorch.