Juan Rocamonde
Juan Rocamonde
Can you share what the `accessTokenVAR()` function does?
Please, provide some more information, so the maintainer of this repository can actually solve your issue. Just providing a console log is not reproducible.
Also found this issue.
Related: #484 (#483). Would be a nice next step after 484 gets merged. There are a few things to polish in the contributors guidelines, especially when it comes to development...
Getting 2534 passed, 588 skipped, 4697 warnings in 126.70s (0:02:06) on my M1 Max when limiting the number of cores to 4. (I think that's how many the medium macOS...
Just ran the flaky test 1000 times and got the following: ``` $ pytest tests/test_examples.py::test_run_example_notebooks --flake-finder --flake-runs=1000 -n 10 -vv >>> [...] 1000 passed, 20311 warnings in 930.04s (0:15:30) ```...
(pulled all commits from #460 so that we can start integrating that into this issue)
Added initial skeleton for the regularization module. Let me know your thoughts. Seems like weight decay has to act after `loss.backward()`, but L2 regularization / arbitrary loss penalties have to...
Sounds good, thanks for doing that. I suspected it was possible but wasn't sure how to. On Mon, Jul 25, 2022 at 7:39 PM, Adam Gleave < ***@***.*** > wrote:...
> Right now `Regularize` seems to be doing two different things: > > * Adaptive regularization via `update_params`, implemented in `WeightDecayRegularizer`. > * Support for regularizing with a variable scale...