init dataset card

README.md

@@ -1,3 +1,20 @@
 ---
 license: mit
 ---
+
+## Dataset Description
+A dataset of pairs of TypeScript code to appropriate type declarations.
+
+## Language
+TypeScript only.
+
+## To Load
+```python
+from datasets import load_dataset
+
+load_dataset("noahshinn024/ts-code2td")
+```
+
+## Distribution of type declaration code lengths
+- uses the tokenizer from [bigcode/santacoder](https://huggingface.co/bigcode/santacoder)
+![](./media/declaration_token_distr.png)

test.py

@@ -0,0 +1,32 @@
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import pandas as pd
+import matplotlib.pyplot as plt
+
+df1 = pd.read_parquet('data/train-pairs.parquet')
+df2 = pd.read_parquet('data/test-pairs.parquet')
+
+tokenizer = AutoTokenizer.from_pretrained('bigcode/santacoder')
+
+df = pd.concat([df1, df2], axis=0)
+df['tokens'] = df['declarations'].apply(lambda x: tokenizer.tokenize(x))
+
+mean_tokens = df['tokens'].apply(lambda x: len(x)).mean()
+max_tokens = df['tokens'].apply(lambda x: len(x)).max()
+min_tokens = df['tokens'].apply(lambda x: len(x)).min()
+num_long_items_2048 = df[df['tokens'].apply(lambda x: len(x) > 2048)].shape[0]
+proportion_2048 = num_long_items_2048 / df.shape[0]
+num_long_items_256 = df[df['tokens'].apply(lambda x: len(x) > 256)].shape[0]
+proportion_256 = num_long_items_256 / df.shape[0]
+
+plt.hist(df['tokens'].apply(lambda x: len(x)), bins=25, range=(0, 2048))
+plt.xlim(0, 2048)
+plt.xlabel('Number of Tokens')
+plt.ylabel('Count')
+plt.title('Distribution of Type Declaration Num Tokens')
+
+# Add a label to the plot with the mean, max, min, and proportion
+label1 = f"Mean: {mean_tokens:.2f}\nMax: {max_tokens}\nMin: {min_tokens}\nProportion > 256: {proportion_256:.2f}\nProportion > 2048: {proportion_2048:.2f}"
+plt.gca().text(0.95, 0.95, label1, transform=plt.gca().transAxes, fontsize=14, verticalalignment='top', horizontalalignment='right')
+
+plt.savefig('declaration_token_distr.png')
+plt.show()