Insn: You should upload simple-baseline.py and describe it in simple-baseline.md. Your simple-baseline.md should say what score your evaluation metric gives to the simple baseline for your test set.
In this project, we attempt two tasks:
- poetry generation
- meter classification
We are attempting to generate poetry to fit a certain meter and rhyme scheme, so a basic baseline would be an n-gram model that generates text character by character. When fed in a bunch of relevant poems of the same meter, the n-gram model would spit out an example baseline poem.
We hope to expand our model to a word-based model (compared to just a character-based one) and to involve neural networks as well to contrast their performance.
Our simple baseline attained a perplexity of 4.614 when run through our evaluation script on perp_poem.txt in /data.
For this milestone, we only focused on classifying whether a line of a poem is in iambic pentameter. Our baseline for this is even more basic than our baseline for the previous task. We randomly output 1 or 0 where 1 means it is in iambic and 0 means it is not.
In the near future, we aim to use the words in the CMU dictionary classified into phonemes and stresses to aid us in classifying a meter. We wish to expand beyond just iambic pentameter to include iambic trimeter and ballad meter as well.
TODO: Our simple baseline attained an f-score of 0.429 when run through our evaluation script on our sample poem for classification class_poem.txt in /data.