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

Fix a couple typos (#1451)

Browse Source 
* Mixtral quantized instruct.

* Fix a couple typos.
This commit is contained in:
Laurent Mazare
2023-12-17 05:20:05 -06:00
committed by GitHub
parent c630622a07
commit 1e86717bf2
24 changed files with 44 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
candle-core/src/quantized/avx.rs
View File

@ -353,7 +353,7 @@ pub(crate) fn vec_dot_q3k_q8k(n: usize, xs: &[BlockQ3K], ys: &[BlockQ8K]) -> Res
q3 = q3.add(32);
// Prepare low and high bits
// We hardcode the shifts here to avoid loading them into a seperate register
// We hardcode the shifts here to avoid loading them into a separate register
let q3l_0 = _mm256_and_si256(q3bits, m3);
let q3h_0 = if j == 0 {
_mm256_srli_epi16(_mm256_andnot_si256(hbits, _mm256_slli_epi16(mone, 0)), 0)
@ -586,7 +586,7 @@ pub(crate) fn vec_dot_q5k_q8k(n: usize, xs: &[BlockQ5K], ys: &[BlockQ8K]) -> Res
let q5bits = _mm256_loadu_si256(q5 as *const __m256i);
q5 = q5.add(32);
//Similar to q3k we hardcode the shifts here to avoid loading them into a seperate register
//Similar to q3k we hardcode the shifts here to avoid loading them into a separate register
let q5l_0 = _mm256_and_si256(q5bits, m4);
let q5l_0_shift_input = _mm256_and_si256(hbits, hmask);
let q5l_0_right_shift = match j {

2
candle-core/src/quantized/gguf_file.rs
View File

@ -463,7 +463,7 @@ impl Content {
) -> Result<QTensor> {
let tensor_info = match self.tensor_infos.get(name) {
Some(tensor_info) => tensor_info,
None => crate::bail!("cannot find tensor-infor for {name}"),
None => crate::bail!("cannot find tensor info for {name}"),
};
tensor_info.read(reader, self.tensor_data_offset)
}