Problem with DataBlock creation for Lesson 8

Part 1 (2020)
1

1. I’m trying to go through Lesson 8 with my own dataset and getting an error on this step:

dls_lm = DataBlock(
    blocks   = TextBlock.from_df(text_cols = 'SUMMARY', is_lm=True, min_freq=2),
    splitter = RandomSplitter(0.1, seed = seed)
)

dls_lm.dataloaders(main_df, bs=16, seq_len=20)

The Error:

TypeError: unhashable type: 'L'

…
2. If I’m trying to add get_x as I can see in Documentation:

dls_lm = DataBlock(
    blocks   = TextBlock.from_df(text_cols = 'SUMMARY', is_lm=True, min_freq=2),
    get_x    = ColReader('SUMMARY'),
    splitter = RandomSplitter(0.1, seed = seed)
)

dls_lm.dataloaders(main_df, bs=16, seq_len=20)

…then error changes to:

AttributeError: 'Series' object has no attribute 'SUMMARY'

What I’m doing wrong, dear community? Any ideas?

2

You should pass in “text”, as the tokenizer moved your tokenized text from your original column to “text”

3

Thank you a lot - it works!