Request for dynamic `num_warmup` argument to MCMC object
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 2.8k
- Forks
- 315
- Avg merge
- 3d 9h
- Merged PRs (30d)
- 27
Description
It would be useful if it wasn't a requirement to specify the number of warmups for MCMC sampling before the model is compiled to save time if it is necessary to re-run the MCMC with more iterations to ensure convergence
Please, could this be changed so that num_warmup is specified after the model has compiled?
(I also started a discussion about this on the pyro forum )
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by tracing the MCMC object's model-compilation and sampling flow, focusing on when num_warmup is currently required. Determine how to allow num_warmup to be supplied after compilation, then verify that rerunning with more warmups does not recompile the model and still supports convergence checks.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100