Max string attribute length exceeded

ERROR: Error occurred during asynchronous operation processing: Max string attribute length (16384) exceeded
Issue: A string or NumPy array that you are logging is too long. Neptune has a character length limit of 16384.
Tip: To help you locate the problem in your code, try initializing Neptune in synchronous mode:
run = neptune.init(
    mode="sync",
)
Workarounds
Text
Split the text into chunks, turn it into a list, and log the list items as a string series:
for epoch in range(epochs_nr):
    token = str(...)
    run["dataset/tokens"].log(token)
Or save the data as a text file and upload that instead:
run["data"].upload("sample_data.csv")
NumPy arrays
Option A) Long arrays are logged with suspension points ("...") instead of all values. You can increase the threshold of NumPy with set_printoptions() and then log it to Neptune as a binary .npy file:
np.set_printoptions(threshold=np.inf)
​
long_array = ...
​
with open('./long_array.npy', 'wb') as f:
    np.save(f, long_array)
​
run = neptune.init()
run["long_array"].upload(./long_array.py)
Option B) Log the array to Neptune as an image:
img_array = ...
run = neptune.init()
run['train/distribution'].log(File.as_image(img_array))
Option C) Log the array as as CSV file.
Neptune's interactive table widget gives you the raw values and an interactive way to sort and filter rows.
for epoch in range(epochs_nr):
    token = str(...)
    run["data/tokens"].log(token)

Troubleshooting - Previous

Waiting for remaining operations

Next - Troubleshooting

Float attribute value malformed

Last modified 1mo ago

Export as PDF

Copy link

Outline

Workarounds

Text

NumPy arrays