[Question] Using bruteforce to retrieve recommendations in Sequential Model

In this example: https://github.com/tensorflow/recommenders/blob/main/docs/examples/sequential_retrieval.ipynb
Could you give an example of how to use BruteForce to retrieve recommendations for sequence data?
when using spotlight by @maciejkula
i can get sequence recommendations from API by entering API body like this
{ "data": [ "End Game", "Age of Ultron", "Iron Man", "SpiderMan No Way Home", "Captain Marvel" ] }
using this function,

def recommend_next_movies(movies, metadata, model, n_movies=15):
   movie_ids = [get_movie_id(movie, metadata) for movie in movies]
   pred = model.predict(sequences=np.array(movie_ids))
   indices = np.argpartition(pred, -n_movies)[-n_movies:]
   best_movie_ids = indices[np.argsort(pred[indices])]
   return [get_metadata(movie_id + 1, metadata) for movie_id in best_movie_ids]

def predict(movies):
        d = recommend_next_movies(movies, metadata, model, n_movies=15)
        df = pd.DataFrame(list(chain.from_iterable(d)))
        return df

Is call function can handles this issue? thank you

any idea how to do this cc @juliobguedes

Start with docs/examples/sequential_retrieval.ipynb and review how the sequential model produces recommendations. Investigate how BruteForce accepts sequence data and compare it with the referenced Spotlight example; done means the notebook documents a working sequence-input retrieval example with expected recommendations.