-
Since the current shape Info for the `Kernel` object is stored inside an `ndarray` in the form like `shape1 = ndarray`, there is currently error like https://github.com/DarrenZhang01/TensorFlow_GSoC/i…
-
```
line 190, in new_kernel_fn
raise ValueError(f'{kernel_fn} requires `{key} == {v_pos}`, '
ValueError: requires `channel_axis == 0`, but input kernel has `channel_axis == 1`, making the infi…
-
In the `v_pos = v % ndim` in the `new_kernel_fn` function, the `ndim` variable is still zero although I have done shape conversion on `k.shape1`. Need to investigate the reason and find out a solution…
-
```
self.assertEqual(x_dict[field.name], y_dict[field.name])
File "/Library/Frameworks/Python.framework/Versions/3.8/lib/python3.8/unittest/case.py", line 912, in assertEqual
assertion_fu…
-
Hi Roman @romanngg , would you mind explaining the use of `jax.lax.cond` in [https://github.com/google/neural-tangents/blob/master/neural_tangents/utils/utils.py#L359-L361](https://github.com/google/n…
-
Hi Roman @romanngg , would you mind explaining a bit about the kernel shape calculation, especially the `_propagate_shape` function? I am also curious about the use of `akey`. Thanks a lot!
[https:…
-
Thanks for your efforts on creating this library! However, when I'm using neural-tangents to compute NTK matrix values, I find it hard to tell the value of kernel_fn for specific inputs, as it's flood…
ybj14 updated
4 years ago
-
When calling `predict` in `nt.predict.gradient_descent` with variables of the following dimensions,
`g_dd [256,256]`
`g_dt [256,256]`
`fx_train [256,1]`
`fx_test [256,1]`
The tuple of predictions…
-
I believe the following line should be `kernel_fn(x_test, x_train, params)` instead of `kernel_fn(x_train, x_test, params)`.
https://github.com/google/neural-tangents/blob/38e9ba906f9cac7a6daa270c2…
-
Hi, could you confirm that the input is `(-1, 7)` in [https://github.com/google/neural-tangents/blob/master/tests/stax_test.py#L1017](https://github.com/google/neural-tangents/blob/master/tests/stax_t…