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

move i2 to quantize

Browse source
This commit is contained in:
root 2024-06-09 18:20:32 +08:00
parent ca09085593
commit dbee0a86c1
3 changed files with 48 additions and 150 deletions
Download patch file Download diff file Expand all files Collapse all files

6
ggml.c
View file

@ -21750,7 +21750,11 @@ size_t ggml_quantize_chunk(
assert(false);
}
GGML_ASSERT(result == nrows * row_size);
if (type == GGML_TYPE_I2) {
result = nrows * row_size / 4 + 32;
} else {
GGML_ASSERT(result == nrows * row_size);
}
return result;
}