WooperNLP solution to Axolotl24 shared task

Notebook running order:

augmentation.ipynb: Augment the dataset to create new example rows to make all senses have at least 3 examples, and create an artificial gloss for each example where gloss is missing, based on the example usage.
get-embeddings.ipynb: Generate the embeddings for a given dataset; for developing the datasets from "dev-testing" folder and for the shared task test, the "augmented" folder. Two different kind of embeddings are generated:

concatenate-embeddings.ipynb: Concatenate both the example and gloss embedding for each row and save them as a new kind of embedding.
axolotl_solution.ipynb: [SUBTASK 1] Given the embeddings (any of the 3 types), a clustering method and a score method for clustering, assign a cluster number (sense_id) for each row.
select_sense.ipynb: [SUBTASK 2] Given the glosses from the augmentation step, select the most proper for each cluster within the clustered rows from subtask 1, as well as prearing the data for sending for both subtasks.

Name		Name	Last commit message	Last commit date
Latest commit History 54 Commits
baselines		baselines
data		data
embeddings		embeddings
evaluation		evaluation
predictions		predictions
.gitignore		.gitignore
README.md		README.md
augmentation.ipynb		augmentation.ipynb
axolotl_solution.ipynb		axolotl_solution.ipynb
concatenate-embeddings.ipynb		concatenate-embeddings.ipynb
create-dev-testing.ipynb		create-dev-testing.ipynb
get-embeddings.ipynb		get-embeddings.ipynb
mix-embeddings.ipynb		mix-embeddings.ipynb
rename-and-prepare.py		rename-and-prepare.py
requirements.txt		requirements.txt
subtask1.md		subtask1.md
subtask2.md		subtask2.md

Provide feedback