dotnet / dotnet/machinelearning
Caching discrepancy between API docs and TrainerInfo.WantCaching property
- Dominant language
- C#
- Stars
- 9.4k
- Forks
- 2k
- Avg merge
- 2d 20h
- Merged PRs (30d)
- 11
Description
There are discrepancies in the docs for trainers on whether caching is required or not.
For example: [MatrixFactorizationTrainer](https://github.com/dotnet/machinelearning/blob/fbd1b93065b451401b1e3276e5ac65b9f303f90b/src/Microsoft.ML.Recommender/MatrixFactorizationTrainer.cs)
In the documentation, it says caching is required:
| Is caching required? | Yes |
However, the `TrainerInfo.WantCaching` property is set to false.
```csharp
_info = new TrainerInfo(normalization: false, caching: false);
```
https://github.com/dotnet/machinelearning/blob/fbd1b93065b451401b1e3276e5ac65b9f303f90b/src/Microsoft.ML.Recommender/MatrixFactorizationTrainer.cs#L370
To-Do:
- [x] Get list of trainers these discrepancies occur.
Contributor guide
Assessment
This issue has not been assessed yet.