Xiao-Yong Jin

Setting `indent-guide-delay` to a value larger than the popup-delay resolve this issue for me.

The behavior of the stateless RNG initializers encumbers reproducibility. As `tf.random.set_seed` no longer affects the initializers, and the only way to have a reproducible initializer is explicit passing a seed...

> The global seed setting for keras initializer is tf.keras.utils.set_random_seed(), which should make all the initializer to have a determined seed. Indeed. Thanks. The doc string https://github.com/keras-team/keras/blob/b80dd12da9c0bc3f569eca3455e77762cf2ee8ef/keras/utils/tf_utils.py#L44-L53 needs to be...

Removed `static_assert` as we discussed. I left comments in the constructors to remind ourselves in the future.

You could also just do ``` diff --git a/ggml.c b/ggml.c index 42e3ee3..80d539f 100644 --- a/ggml.c +++ b/ggml.c @@ -2216,7 +2216,10 @@ static void ggml_vec_dot_q4_0(const int n, float * restrict s,...

Let us know if you need any kind of help.

Try starting with temperature `T=10`, so that the two Gaussians are only 4 σ apart.

Try raising the temperature for +/-4. Keep two gaussians 4 sigma apart, which means start with temperature 40.

Alpaca RLHF is for instruction only. Make it interactive doesn't really make sense. It would be simpler just to have two different command line arguments tailored for instructions with or...

Tried https://github.com/tloen/alpaca-lora on a 13B model from Hugging Face. This is the diff for alpaca-lora, ``` diff diff --git a/export_state_dict_checkpoint.py b/export_state_dict_checkpoint.py index 78e9d1f..3b88cb9 100644 --- a/export_state_dict_checkpoint.py +++ b/export_state_dict_checkpoint.py @@ -11,10...