huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-18 19:47:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

More model cloning. (#1126)

Browse Source 
* More model cloning.

* More cloning on quantized models.
This commit is contained in:
Laurent Mazare
2023-10-18 21:55:46 +01:00
committed by GitHub
parent 185b54a33b
commit 902d0b9166
4 changed files with 19 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
candle-transformers/src/models/quantized_mpt.rs
View File

@ -7,7 +7,7 @@ use candle_nn::LayerNorm;
pub use super::mpt::Config;
#[derive(Debug)]
#[derive(Debug, Clone)]
struct GroupedQueryAttention {
wqkv: Linear,
out_proj: Linear,
@ -101,7 +101,7 @@ impl GroupedQueryAttention {
}
}
#[derive(Debug)]
#[derive(Debug, Clone)]
struct Ffn {
up_proj: Linear,
down_proj: Linear,
@ -122,7 +122,7 @@ impl Module for Ffn {
}
}
#[derive(Debug)]
#[derive(Debug, Clone)]
struct MPTBlock {
norm1: LayerNorm, // Do we need the low-precision variant?
attn: GroupedQueryAttention,
@ -155,7 +155,7 @@ impl MPTBlock {
}
}
#[derive(Debug)]
#[derive(Debug, Clone)]
pub struct Model {
wte: Embedding,
blocks: Vec<MPTBlock>,