gridfm / gridfm/gridfm-graphkit
use .pt files instead of whole parquet datasets for normalization and AC/DC evaluation
Open
Nobody has claimed this yet.
enhancement
- Dominant language
- Python
- Stars
- 105
- Forks
- 36
- Avg merge
- 1d 8h
- Merged PRs (30d)
- 9
Description
Why?
- More consistent with the rest of the code
- We can then share only the .pt files instead of having to share both
- saves RAM by just loading the samples we need to norm fit instead of all the dataset
- Speeds up the evaluation of the AC/DC solvers on the test set
- Allows fitting normalizer and AC/DC eval without using 1TB or RAM for case10k
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
The issue does not name files, tests, or entry points; first locate the normalization workflow, AC/DC evaluation commands, and their parquet-loading code. Trace how datasets are loaded and identify the existing .pt data path, if any. Done means normalization and test-set evaluation use the needed .pt samples without loading whole parquet datasets or requiring excessive RAM.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- data, machine-learning
- Issue type
- Refactor
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100