nicolas
/
ik_llama_opt
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix MXFP4 non-interlevaed type (#2306)

This commit is contained in:
Kawrakow 2026-08-13 08:41:39 +02:00 committed by GitHub
parent 79c1e16a41
commit ee77f7ffb8
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/llama-quantize.cpp
View File

@ -110,7 +110,7 @@ std::pair<ggml_type, int> interleaved_properties(ggml_type type) {
{ GGML_TYPE_IQ4_KS_R4, { GGML_TYPE_IQ4_KS, 4} },
{ GGML_TYPE_IQ5_KS_R4, { GGML_TYPE_IQ5_KS, 4} },
{ GGML_TYPE_IQ5_K_R4, { GGML_TYPE_IQ5_K, 4} },
{ GGML_TYPE_MXFP4_R8, { GGML_TYPE_MXFP4_R8, 8} },
{ GGML_TYPE_MXFP4_R8, { GGML_TYPE_MXFP4, 8} },
{ GGML_TYPE_Q8_KV_R8, { GGML_TYPE_Q8_KV, 8} },
{ GGML_TYPE_Q8_K_R8, { GGML_TYPE_Q8_0, 8} },
{ GGML_TYPE_BF16_R16, { GGML_TYPE_BF16, 16} },