Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

WeightWatcher.analyze() not working for wav2vec2

See original GitHub issue

*Perhaps relate to https://github.com/CalculatedContent/WeightWatcher/issues/105

For testing purposes, I’ve fed the wav2vec2 model downloaded from huggingface to the weightwatcher. It is stopped by an error very fast.

The code used is below:

import weightwatcher as ww
from transformers import AutoModelForCTC

model = AutoModelForCTC.from_pretrained("facebook/wav2vec2-base-960h")
watcher = ww.WeightWatcher(model=model)
details = watcher.analyze()
summary = watcher.get_summary(details)

It yields: WARNING:weightwatcher:not enough eigenvalues, stopping Traceback (most recent call last): File “t.py”, line 6, in <module> details = watcher.analyze() File “/home/pete/.pyenv/versions/wristband/lib/python3.8/site-packages/weightwatcher/weightwatcher.py”, line 1649, in analyze self.apply_analyze_eigenvectors(ww_layer, params=params) File “/home/pete/.pyenv/versions/wristband/lib/python3.8/site-packages/weightwatcher/weightwatcher.py”, line 2931, in apply_analyze_eigenvectors X = np.matmul(W.T, W) ValueError: matmul: Input operand 1 has a mismatch in its core dimension 0, with gufunc signature (n?,k),(k,m?)->(n?,m?) (size 1 is different from 512)

Versions: torch==1.12.1+cpu weightwatcher==0.5.6 transformers==4.22.1

Please let me know if you need additional information.