Enrico Fini
Enrico Fini
Yes, your curves look similar to mine. I think it is likely to be due to hyperparam tuning of the offline linear eval. Also, always remember that there might be...
One last thing that just came to my mind. We recently found that Pillow-SIMD can have a detrimental effect on some models (see the issue here https://github.com/vturrisi/solo-learn/issues/313). I am not...
How much did you get with online linear eval?
Ok, so around 57. The checkpoint that I shared should have online eval accuracy 58.8%.
Hi! I have just checked my logs and I don't seem to have saved that experiment. I found the imagenet100 run with random features instead, and it reached 15.34% after...
if the backbone is randomly initialized and frozen there is no difference between online and offline (except augmentations that should not matter much).
Hello! From the paper: > LUMP uses k-NN evaluation, therefore we adapt the code provided by the authors to run in our code base. For Lin et al., we compare...
IIRC all the other parameters were left as per their default value in the respective paper/code
Hello, - I used reservoir sampling as well - I also used color jittering (basically SimCLR augmentations)
Hi! I am not sure I understand the issue. Table C reports results using offline k-nn, while I think you are running online k-nn, am I right? Table 4 is...