Closed vturrisi closed 2 years ago
Merging #275 (bbc2b46) into main (0273243) will decrease coverage by
1.06%
. The diff coverage is42.03%
.
Flag | Coverage Δ | *Carryforward flag | |
---|---|---|---|
cpu | 82.09% <45.71%> (-1.01%) |
:arrow_down: | |
dali | 40.66% <11.76%> (ø) |
Carriedforward from 6de7de3 |
*This pull request uses carry forward flags. Click here to find out more.
Impacted Files | Coverage Δ | |
---|---|---|
solo/utils/__init__.py | 83.33% <ø> (+2.08%) |
:arrow_up: |
solo/utils/dali_dataloader.py | 37.96% <0.00%> (+0.16%) |
:arrow_up: |
solo/utils/h5_dataset.py | 23.88% <23.88%> (ø) |
|
solo/utils/pretrain_dataloader.py | 58.79% <33.33%> (-1.80%) |
:arrow_down: |
solo/utils/classification_dataloader.py | 47.22% <41.66%> (-6.81%) |
:arrow_down: |
solo/methods/base.py | 78.40% <50.00%> (-0.32%) |
:arrow_down: |
solo/utils/misc.py | 68.18% <50.00%> (-1.89%) |
:arrow_down: |
solo/args/dataset.py | 76.19% <60.00%> (+1.19%) |
:arrow_up: |
solo/methods/linear.py | 88.83% <66.66%> (+0.37%) |
:arrow_up: |
solo/args/utils.py | 69.93% <80.00%> (+0.85%) |
:arrow_up: |
... and 11 more |
Hello @vturrisi and @DonkeyShot21,
Could I ask what is benefit or difference between using data h5 and the traditional imagenet folder?
hey @trungpx, you can read about it here: https://www.h5py.org/
We are still missing the h5 file constructor, the rest is in place.