Project-MONAI / Project-MONAI/model-zoo
Allow for multiple experiments and runs out of the box
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 338
- Forks
- 106
- PR merge metrics
- No merged PRs in 30d
Description
Is your feature request related to a problem? Please describe.
Monai bundle should allow for multiple experiments and runs out of the box
Describe the solution you'd like
In the train.config
- Add "folder_name": "experiment1"
this should lead to - creating a new folder in models/experiment1
- Make new model, tensor board etc be generated in this new folder
- Copy the train.json and train_continues.json into this folder for tracking
- direct the log file to be generated into this new folder
Describe alternatives you've considered
I have modified the bundle myself to do some of this. however I need to give the same folder name in the train and in the logging.config
Additional context
the infer.config should be modified to take parameters from the train.json files specifically the intenisty range , resolution etc
I couldn't get this to work. I tired
- passing in the train.json gave errors
- changed parameters in the infer.json to use the filepath#parametername didn't work as the train.json is in subfolder model/experiment1
- copying the infer.json into the new folder model/experiment1 but I got errors related to path
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 how train.config, logging.config, train.json, and train_continues.json are handled when a bundle runs. Review the infer.config/infer.json parameter references and the model/experiment1 path behavior. Done means multiple runs create an experiment folder containing models, TensorBoard output, logs, and copied training configs, while inference can use the recorded training parameters.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100