Open rusheb opened 1 year ago
I'm not really sure what's wrong here. It may be a 3.11 only issue. I haven't had a chance to test myself
also cannot run command listed in README on linux, though my error is different:
$ elk elicit microsoft/deberta-v2-xxlarge-mnli imdb
Traceback (most recent call last):
...
File "/mnt/md1/nets/elk/elk/extraction/extraction.py", line 336, in extract
raise ValueError("Can only extract LM predictions from autoregressive models.")
ValueError: Can only extract LM predictions from autoregressive models.
update: I found replacing the model with gpt2
worked fine, after a minor tweak to fix an exception name issue.
Summary
Hi there,
I was attempting to set up the application based on the Quick Start guide in the README. When trying to run the
elk elicit microsoft/deberta-v2-xxlarge-mnli imdb
command, the program terminates with a ValueError.Can you please help me understand what's going wrong and how I can fix it?
Thanks!
Steps to reproduce
Environment
Commands run
I also tried a couple of suggestions from @KayKozaronek, and got the same error:
elk elicit gpt2 boolq
--net
arg:elk elicit microsoft/deberta-v2-xxlarge-mnli imdb --net ccs
Error output
(Full stack trace at the bottom of this post)
Mitigation
GPT tells me we should fix this by replacing the following property from
sweep.py
with adefault_factory
, but without more context on the project I can't tell whether this is the right solution.Full stack trace