Update README.md

README.md

@@ -25,8 +25,13 @@ You can find the paper here: https://arxiv.org/abs/2311.10770, and the abstract
 
 This is the raw pretraining checkpoint. You can use this to fine-tune on a downstream task like GLUE as discussed in the paper. This model is provided only as sanity check for research purposes, it is untested and unfit for deployment.
 
-### How to use
+### How to get started
 
+1. Create a new Python/conda environment, or simply use one that does not have any previous version of the original `cramming` project installed. If, by accident, you use the original cramming repository code instead of the one provided in the `/training` folder of this project, you will be warned by `transformers` that there are some extra weights (FFF weight) and that some weights are missing (the FF weights expected by the original `crammedBERT`).
+2. `cd ./training`
+3. `pip install .`
+4. Create `minimal_example.py`
+5. Paste the code below
 
 ```python
 import cramming
@@ -40,6 +45,8 @@ encoded_input = tokenizer(text, return_tensors='pt')
 output = model(**encoded_input)
 ```
 
+6. Run `python minimal_example.py`.
+
 
 ### Limitations and bias