Kevin P Murphy
Kevin P Murphy
Another example from ch 2 (probability)
I noticed this problem, too: I cannot get the SGD optimizer in Pytorch to overfit for the degree-20 polynomial, even when using full-batch training. (I did not try exact least...
IIUC, it should not be necessary to install jax or jaxlib on colab, since it is built in. See eg [this lenet_jax](https://github.com/probml/pyprobml/blob/master/notebooks/book1/14/lenet_jax.ipynb) notebook.
how do I make this happen in latex?
ok, i've added this, it will appear in the next release (later tonight :)
[This colab](https://colab.research.google.com/drive/19VrDpEuuhkWUNGQkx3dXcW08DeuuH-t2?usp=sharing) makes it clear that the problem does not exist in 0.16 (colab default) but does exist in 0.18.
Hi @SiegeLordEx Yes, that fixes it - thanks!
I get this error ``` pip install elegy==0.8.6 ERROR: Ignored the following versions that require a different python version: 0.6.0 Requires-Python >=3.6.1,=3.6.1,=3.6.1,=3.6.1,=3.6.1,=3.7,=3.7,=3.7,=3.7,=3.7,=3.7,=3.7,=3.7,=3.7,=3.7,
When I don't pin the version number I get very old versions of everything ``` pip install elegy Collecting elegy Using cached elegy-0.5.0-py3-none-any.whl (206 kB) Collecting jaxlib=0.1.59 Using cached jaxlib-0.1.76-cp310-none-manylinux2010_x86_64.whl...
Atfer installing this very old version it does not run , presumaly because it relies on an old tpu driver. (I am running on a TPU VM V3-8) ``` base)...