huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-18 11:37:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix dequantization. (#1823)

Browse Source
This commit is contained in:
Laurent Mazare
2024-03-08 23:12:13 +01:00
committed by GitHub
parent 3440cec3a0
commit 936f6a4840
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@ -398,7 +398,7 @@ impl QMatMul {
_ => DEQUANTIZE_ALL.with(|b| *b), _ => DEQUANTIZE_ALL.with(|b| *b),
}; };
let t = if dequantize { let t = if dequantize {
let tensor = qtensor.dequantize(&Device::Cpu)?; let tensor = qtensor.dequantize(&qtensor.device())?;
Self::Tensor(tensor) Self::Tensor(tensor)
} else { } else {
Self::QTensor(qtensor) Self::QTensor(qtensor)