-
I first wrongly created an issue for julia itself, but it is apparently an issue related to SparseArrays.jl. you can find the previous issue [here](https://github.com/JuliaLang/julia/issues/56163#issu…
-
Are you expecting this? It's platform-specific, this is for the M1, Linux is different IIRC.
``` r
jsonlite::toJSON(-0x0p+0)
#> [-0]
-0x0p+0 == 0
#> [1] TRUE
```
Created on 2024-02-03 with …
-
Dear LLMC team,
I've been trying to run mixed-precision PTQ quantization using RTN.
I suspect there's a bug, as the **non-default settings in `mix_bits` are ignored**.
My understanding of the cod…
-
Versions:
- CUDA 12.2
- Debian 11
- triton==3.1.0
- torch==2.1.2
I have a kernel where in one particular line I define a tensor of -inf. This is all in the body of a function decorated with tri…
-
**Describe the bug**
If you import zip code data into an extra_data column and set that column's data type to `text`, the data will appear in the json data correctly, but when requesting the data on …
-
It turns out FGDB featureclass's attribute table won't allow field names that begin with a number. Please move the year numbers behind alphabetical characters. This modification will greatly improve t…
jdduh updated
2 weeks ago
-
## 📝 Description
Given an array arr. Return the modified array in such a way that if the current and next numbers are valid numbers and are equal then double the current number value and replace the…
-
## Detailed Description
Check for zeros when loading the data. This should only be used when running inference
## Context
- good to check for lots of zeros and fail hard
## Possible Implem…
-
There is a bug in the current AD implementation for eigsolve when the eigenvalues are degenerate.
A simple example
```julia
using KrylovKit
using Zygote
@testset "degenerated eigsolve ad" …
-
It looks like in the implementation it allows number literals to have leading zeros (i.e. `09` results in `9`). This is valid per the grammar in the spec, but it doesn't match the description:
> Nu…