vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Remove Q4_3 which is no better than Q5 (#1218)

Browse source
This commit is contained in:
Stephan Walter 2023-04-28 23:10:43 +00:00 • committed by GitHub
parent 7f15c5c477
commit 36d19a603b
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
11 changed files with 21 additions and 359 deletions
Download patch file Download diff file Expand all files Collapse all files

2
llama.h
View file

@ -73,7 +73,7 @@ extern "C" {
LLAMA_FTYPE_MOSTLY_Q4_1 = 3, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
LLAMA_FTYPE_MOSTLY_Q4_2 = 5, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q4_3 = 6, // except 1d tensors
// LLAMA_FTYPE_MOSTLY_Q4_3 (6) support has been removed
LLAMA_FTYPE_MOSTLY_Q8_0 = 7, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q5_0 = 8, // except 1d tensors
LLAMA_FTYPE_MOSTLY_Q5_1 = 9, // except 1d tensors