Update README.md

README.md

@@ -1,6 +1,7 @@
 ---
 license: apache-2.0
 datasets:
+- Skylion007/openwebtext
 - rajpurkar/squad
 - google-research-datasets/natural_questions
 - hotpotqa/hotpot_qa
@@ -11,13 +12,13 @@ pipeline_tag: question-answering
 - Project Type: Bring your own method
 
 ## Structure
-- `data/` contains the data used for the project (after running `load_data.py`, and downloading the natural questions)
-- `distilbert.py` contains the code for the DistilBERT model and the Dataset. A function for testing the functionality is in there too.
+- `data/` contains the data used for the project
+- `distilbert.py` contains the code for the DistilBERT model and the Dataset.
 - `distilbert.ipynb` contains the creation and training of the DistilBERT model
 - `distilbert.model` is the distilbert model
 - `distilbert_reuse.model` is the question answering model
-- `load_data.py` contains the code for loading the data and preprocessing it. We also split it up into smaller files to load in the Dataset later on.
-- `qa_model.py` contains the code for thee different QA models. We also define a separate Dataset class in there and a method for testing the models.
+- `load_data.py` contains the code for loading the data and preprocessing it.
+- `qa_model.py` contains the code for thee different QA models.
 - `qa_model.ipynb` contains the creation and training of the QA models.
 - `requirements.txt` contains the requirements for the project
 - `utils.py` contains some helper functions for the project. It contains the functions to evaluate the models and a way to visualise the trained parameters for each model.
@@ -63,7 +64,7 @@ Now for the Question Answering model.
 * Target for Error Metric:
   * EM: 0.6
   * F-1: 0.7
-* Achieved value: I almost achieved the target for both of the measurements. I ultimately quit I had already spent a lot of time on the project and thought that the results were reasonable.
+* Achieved value: I almost achieved the target for both of the measurements.
   * EM: 0.52
   * F-1: 0.67