HatefulMemesScenario get_instances returning error

[dxwu2]

Calling get_instances for HatefulMemesScenario is getting me the following:
Exception: Failed with exit code 8: wget 'https://huggingface.co/datasets/neuralcatcher/hateful_memes/resolve/main/img/23058.png?download=true' -O output/benchmarks/downloaded/hatefulmemes/img/23058.png.tmp

Specifically, the traceback is:

• line 66 in common/general.py, in shell: raise Exception(f"Failed with exit code {exit_code}: {cmd}")
• line 94 in common/general.py, in ensure_file_downloaded: shell([downloader_executable, source_url, "-O", tmp_path])

[yifanmai]

Could you provide the complete error logs, including the exception message itself?

cc @teetone

[dxwu2]

[yifanmai]

OK, I can replicate this locally. There is indeed a row for image 23058.png which is missing from this Hugging Face mirror.

@teetone How are you running this in production? Are you using the canonical version of the dataset instead?