bion howard
bion howard
does this mean we need to change the button props after the first signin?
Is the preferred usage right now something like ``` struct Bar { baz: String } let bar = Bar { baz: "qux".to_owned() }; let bar_string = serde_json::to_string(&bar)?; redis::cmd("JSON.SET").arg("foo").arg(".").arg(bar_string).query(con)?; ``` Do...
this would make the code simpler, less `if env_name in unity_env_names: unity_env = UnityEnvironment(PATH_LOOKUP[env_name])`
truly amazing
@bayerj @sguada bump, I'd love to use gin to configure Optuna or Keras-Tuner searches; i have a bunch of bricks each with parameters pulled to the top of the file...
right, M can be feature distance, C1, C2 could be pairwise distances between atoms What I'm wondering about is, can we maintain a last dimension of the data? instead of...
The main issue for our use case we run out of GPU memory. KeOps exists to solve that problem by only doing part of the problem on the GPU at...
whoa, that struct.dataclass is cool, and would solve headaches of passing modules to functions and getting `not a JAX type` errors
yes this makes sense and should be a key feature to not repeat trials unless we've already tried everything and just want to reduce variance
```jl function show_box_space(s) shape = s.shape dtype = pystr(s.dtype) target_type = occursin("int", dtype) ? Int : Float high = map(target_type, Array(s.high)) low = map(target_type, Array(s.low)) println("Box$shape") @show dtype @show high[1]...