huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-19 03:54:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

Quantized support for f16 and f32 (#457)

Browse Source 
* Add f32 as a quantized type.

* Add f16 as a quantized type too.
This commit is contained in:
Laurent Mazare
2023-08-15 21:09:37 +01:00
committed by GitHub
parent e68b2accb4
commit b8263aa15c
3 changed files with 75 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
candle-core/src/cpu/neon.rs
View File

@ -69,9 +69,6 @@ impl Cpu<ARR> for CurrentCpu {
for i in 0..ARR / 4 {
x[4 * i] = vaddq_f32(x[4 * i], x[4 * i + 2]);
}
for i in 0..ARR / 8 {
x[8 * i] = vaddq_f32(x[8 * i], x[8 * i + 4]);
}
*y = Self::reduce_one(x[0]);
}
}