furyhawk/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-07-21 10:15:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add Q2_0 quantization: type definition and CPU backend (#24448)

Browse Source
This commit is contained in:
Pasha Khosravi
2026-07-07 12:05:47 -07:00
committed by GitHub
parent c198af4dc2
commit bec4772f6a
17 changed files with 262 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test-quantize-fns.cpp
+2 -1
View File
@@ -158,6 +158,7 @@ static int test_vec_dot_q(bool verbose) {
type == GGML_TYPE_Q1_0 ? MAX_QUANTIZATION_TOTAL_ERROR_BINARY :
type == GGML_TYPE_TQ1_0 ? MAX_QUANTIZATION_TOTAL_ERROR_TERNARY :
type == GGML_TYPE_TQ2_0 ? MAX_QUANTIZATION_TOTAL_ERROR_TERNARY :
type == GGML_TYPE_Q2_0 ? MAX_QUANTIZATION_TOTAL_ERROR_TERNARY :
type == GGML_TYPE_Q2_K ? MAX_QUANTIZATION_TOTAL_ERROR_2BITS :
type == GGML_TYPE_IQ2_S ? MAX_QUANTIZATION_TOTAL_ERROR_2BITS :
type == GGML_TYPE_Q3_K ? MAX_QUANTIZATION_TOTAL_ERROR_3BITS :
@@ -183,7 +184,7 @@ static int test_vec_dot_q(bool verbose) {
? MAX_DOT_PRODUCT_ERROR_LOWBIT
: type == GGML_TYPE_Q1_0
? MAX_DOT_PRODUCT_ERROR_BINARY
: type == GGML_TYPE_TQ1_0 || type == GGML_TYPE_TQ2_0
: type == GGML_TYPE_TQ1_0 || type == GGML_TYPE_TQ2_0 || type == GGML_TYPE_Q2_0
? MAX_DOT_PRODUCT_ERROR_TERNARY
: type == GGML_TYPE_NVFP4
? MAX_DOT_PRODUCT_ERROR_FP4