Mohamad Amin Mohamadi

I have this problem too! The adding animation works but removing does not. I'm using RecyclerView version 22.2.1

Well I think it is solved for me now. The problem was that I was calling `notifyDataSetChanged()` right after `adapter.remove(position)` and it was cancelling the animation. My solution was to...

Update: I tried layer-wise jacobians: ```python func, params, buffers = make_functional_with_buffers(model) params = list(params) def jac_p(p, i): params[i] = p out = func(params, buffers, input_dict)['logits'] return out J0 = jacrev(jac_p)(params[0],...

Another update: ```python func, params, buffers = make_functional_with_buffers(model) params = list(params) def jac_p_o(p, i, o): params[i] = p out = func(params, buffers, input_dict)['logits'] return out[o] J0 = jacrev(jac_p_o)(params[0], 0, 0)...

Hey, I'm trying to compute this:  Which requires jacobians of `f(X)` with respect to the parameters: `\nabla_\theta f(X)` where `X` is our datapoints. So in a batched manner, it...

This is my pip info: -------------------- Name: torch Version: 1.10.0 Name: functorch Version: 0.0.1a0+6107f49 Should I be using the PyTorch nightly build? > > Actually, even your example gives me...

Oh! Sorry about that. updated the permission! @Chillee

@zou3519 That's nice! thanks for the effort. So what we're computing here, is the "empirical" ntk, which I think was first derived (or shown to be useful) here: https://arxiv.org/pdf/1902.06720.pdf. Basically,...

Hey, Great, looking forward to it! I don't think any of these two shapes can be called the "correct" shape. In fact, they're both used in literature (both empirical and...

I guess the question is, is XLA's `triangular_solve` operator really working in-place? If not, this could be expected, but shouldn't it work inplace?