From 7d30465864358f05aadd83691a0aa9343c44b1b4 Mon Sep 17 00:00:00 2001 From: David Hoffmann Date: Mon, 22 Apr 2024 11:46:41 +0200 Subject: [PATCH] Add folder structure note to readme --- mlprank/architectures/README.md | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/mlprank/architectures/README.md b/mlprank/architectures/README.md index 79c4fab..90fc649 100644 --- a/mlprank/architectures/README.md +++ b/mlprank/architectures/README.md @@ -19,7 +19,8 @@ To add a new model based on a simple MLP architecture, follow the steps below: 3. Add the dataset the model was trained on to the `DATASETS` dictionary in `registry.py` with the following structure: `":{"wrapper":, "path": }`. 4. Train the model as described in the [Tuning and Training MLP Models](https://gitlab.aws.dev/adavidho/mlp-rank-pruning/-/blob/main/training/README.md?ref_type=heads) documentation. 5. Add the path to trained model weights as an attribute to the `MLP_MODEL` dictionary in `registry.py` with the following structure: `"path": `. -6. Lastly, add an entry to `MODEL_DATA` mapping the model name to the name of the dataset it was trained on as they were referenced in `MLP_MODEL` and `DATASETS` respectively. +6. Make sure you already create the folder structure as specified in `registry.py` such that data and models can be saved there. +7. Lastly, add an entry to `MODEL_DATA` mapping the model name to the name of the dataset it was trained on as they were referenced in `MLP_MODEL` and `DATASETS` respectively. ## Adding a Scoring Function