Closed RylanSchaeffer closed 1 year ago
I rm -rf *
my .result_caching
path and the error is gone. Bizarre :thinking:
The error is back :/
The result_caching
library caches results so you don't have to rerun costly computations too many times. My guess is that the architecture identifier architecture:RF-100-cosine-bernoulli-b-ns|task:None|kind:Rand|source:RS|lyr:mlp|agg:pca|n_comp:1000
is not properly written to disk or it has trouble reading the file. I don't think this has anything to do with V4 vs IT but rather the order of execution.
You can disable result_caching
altogether by running with RESULTCACHING_DISABLE=1
or only specific components, e.g. RESULTCACHING_DISABLE=model_tools.activations
or chance the model identifier.
I opened another issue on results_caching
(see https://github.com/brain-score/result_caching/issues/16). It seems the errors appear with low probability if many SLURM jobs are running simultaneously.
I have a script which fits models against Majaj IT and Majaj V4. IT runs fine, but when I try specifying V4 instead, I receive the following stack trace and error:
I'm not familiar with
result_caching
. Could someone please help me understand why this problem emerges for V4 but not IT? What's the solution to fixing it?