-
Hi again,
I'm sorry, but there is another issue which I thought to mention in a different post. After resolving the method errors following your suggestions in the previous Issue post, the computati…
-
Currently, in [Q4_0](https://github.com/ggerganov/ggml/pull/27) quantization we choose the scaling factor for each 32 group of weights as `abs(max(x_i))/7`. It is easy to see that this is suboptimal.
…
-
Kind of an annoyance for C projects is the fact that C++ code is full of references to things which in fact end up requiring their own support libraries. Exceptions, for example, require a support lib…
-
Discussing the memory-related properties of references: does `&T` have to point to allocated memory (with at least `size_of::()` bytes being allocated)? If yes, does the memory have to contain data t…
-
At the moment GenericFFT.jl also exports `fft` from FFTW.jl, such that type promotions take place when using it
```julia
julia> using GenericFFT, SoftPosit
julia> a = Posit16.(randn(8))
8-elemen…
-
## Summary
Thanks to @PhilMiller for an excellent discussion about this! I've written some text to add to the introduction, but we also need normative wording changes. The essence is to replace u…
-
At the risk of opening a horrible can of worms, I'm opening this discussion to get thoughts, opinions, and feedback on the possibility of extending librosa to support just intonation where appropriate…
-
What should each of the following evaluate to?
```js
#[+0] == #[-0];
#[+0] === #[-0];
Object.is(#[+0], #[-0]);
#[NaN] == #[NaN];
#[NaN] === #[NaN];
Object.is(#[NaN], #[NaN]);
```
…
-
https://twitter.com/ProjectPhysX/status/1444007905058070531?s=20 suggests a new conversion between Float32 and Posit16
-
With the posit number system we added a conditionally compiled exception mechanism to enable the use case to trap a piece of code and catch any arithmetic errors, such as division by zero. This initia…