Yuxuan Jiang
Yuxuan Jiang
Thanks for this excellent library! When using `BacksolveAdjoint` adjoint method, it's easy to get (and log) forward pass stats with ```python sol = diffrax.diffeqsolve( terms, solver, t_0, t_T, initial_step_size, init_y,...
In a haiku module, I would like to create N networks with the same structure but different parameters, then apply the same input to the N networks to get a...
Not sure if it is the correct place to ask, but since [drone-docker](https://github.com/drone-plugins/drone-docker) does not allow filing issues, I will post my issue here. https://github.com/drone-plugins/drone-docker/blob/7888a798b1b331bd9c5a2662e50494f907499f2c/card.go#L75-L88 The `mapRegistryToURL` function uses a...
I have to admit that I do not fully understand the necessity of `hk.vmap` instead of `jax.vmap`. Nevertheless, when I need to vmap something, I would use `hk.vmap` whenever the...
Thanks for this awesome library. It's fast, powerful, and has already helped me a lot. The sklearn KMeans algorithm has a optional `sample_weight` option, specifying weights for each observation. I'm...
For example ``` pub fn highlight(r#type: i32); ``` leads to ``` void ffi$cxxbridge1$highlight(::std::int32_t r#type) noexcept { ... } ``` I wish it could be unraw-ed if being a valid C++...
The model becomes instable after many steps, even with simplest control. Repro as follows. Could you investigate the cause? Or suggest workarounds to make the model more robust? ```python import...
Inside containers hosted on a server with a lot of cores and RAM, with the actual core and RAM limit constrained via cgroup settings (cpuset and memory), the htop's CPU...
Currently, unless I am missing something, for an `eqx.Module`, the instance field annotations and instance method annotations are disjoint. This prevents the module input from being sufficiently checked (see 1...