zhanghang1989 / zhanghang1989/PyTorch-Encoding

Training time seems to be too long.

Open
#191 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
2k
Forks
448
PR merge metrics
No merged PRs in 30d

Description

Hi! I am using your code to train an enc-net based on resnet101 on ADE dataset, on 4x1080ti, the batch size is set to 8 with synchronous BN turned on. And I notice that it takes around 2 hours for an epoch, which means that it will cost about 10 days to train this model for 120 epochs(the suggested epoch nums for ADE in your paper). I wonder whether the training time ought to be that long? If not, what's a reasonable time to train this model?

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reviewing the training configuration described for the enc-net with resnet101 on ADE, including batch size 8 and synchronous BN on 4x1080ti GPUs. Done means determining whether the reported two-hour epoch and ten-day 120-epoch estimate are expected, and documenting a reasonable training-time expectation.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
computer-vision, machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.