Bug: Test and training data the same in examples/Customizing_embeddings.ipynb

[dpsimpson]

I'm pretty sure there's a typo in the optimize_matrix routine in block 10 of the customizing_embeddings notebook.

The code

e1_train, e2_train, s_train = tensors_from_dataframe(
        df[df["dataset"] == "train"], "text_1_embedding", "text_2_embedding", "label"
    )
    e1_test, e2_test, s_test = tensors_from_dataframe(
        df[df["dataset"] == "train"], "text_1_embedding", "text_2_embedding", "label"
    )

should probably be

e1_train, e2_train, s_train = tensors_from_dataframe(
        df[df["dataset"] == "train"], "text_1_embedding", "text_2_embedding", "label"
    )
    e1_test, e2_test, s_test = tensors_from_dataframe(
        df[df["dataset"] == "test"], "text_1_embedding", "text_2_embedding", "label"
    )

[ted-at-openai]

Wow. Great catch. I'll fix this.

[ted-at-openai]

Fixed here: https://github.com/openai/openai-cookbook/pull/236