I am trying to train an ULMFiT model found https://github.com/kheyer/Genomic-ULMFiT
I am using custom data and following one of the notebook examples, specifically Mammals LM 3 3m1s. I use the same exact code but just altered to use my own data. When I run a specific line of code
<code>data = GenomicTextLMDataBunch.from_df(path, df[:100000], df_valid, bs=800, tokenizer=tok,
chunksize=50000, text_cols=0, label_cols=1, max_vocab=80000)
</code>
<code>data = GenomicTextLMDataBunch.from_df(path, df[:100000], df_valid, bs=800, tokenizer=tok,
chunksize=50000, text_cols=0, label_cols=1, max_vocab=80000)
</code>
data = GenomicTextLMDataBunch.from_df(path, df[:100000], df_valid, bs=800, tokenizer=tok,
chunksize=50000, text_cols=0, label_cols=1, max_vocab=80000)
I receive an error
<code>ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (47613,) + inhomogeneous part.
</code>
<code>ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (47613,) + inhomogeneous part.
</code>
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (47613,) + inhomogeneous part.
Does anyone have any experience with this model or can try to help?
Much appreciated and thanks in advance!