Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Related Issues
Proposed Changes:
Remove dill dependency and remove dill from from saving models via torch.save with pickle_module set to dill.
How did you test it?
Notes for the reviewer
dill was introduced in FARM in deepset-ai/FARM#271 At that time we used torch.save to not only store the state dicts but also the model itself. As the AdaptiveModel allows defining a loss function as a callable, it was not serializable out of the box and we introduced dill to save it.
In Haystack, we're not storing the model class though and thus I believe dill is not necessary anymore. What remains to be checked is whether model saved with dill can be loaded without. If it's compatible, we can remove dill otherwise we need to keep it.
Checklist
fix:
,feat:
,build:
,chore:
,ci:
,docs:
,style:
,refactor:
,perf:
,test:
.