kwea123 / kwea123/nerf_pl

Training doesn't work / converge for Blender - chair dataset

Open
#136 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Jupyter Notebook
Stars
2.8k
Forks
458
PR merge metrics
No merged PRs in 30d

Description

Hi, I followed the install instructions and tried training for the chair data in Blender. However, it never converged and just output white background. The PNSR just bounced around between 13-14 for 13 epochs. Did anybody else experience something similar with this dataset or others? Thanks!

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reproducing the reported training run from the install instructions with the Blender chair dataset, recording the PSNR and rendered output across epochs. Compare the run configuration and dataset processing with any available examples; done means identifying a reproducible cause for the non-convergence and documenting a resolution or confirmed dataset limitation.

Written by the indexing model from the issue text.

Assessment

Tech stack
jupyter-notebook, pytorch
Domain
machine-learning
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.