diff options
| author | zmlcc <zmlcc@outlook.com> | 2023-09-08 20:29:40 +0800 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-09-08 13:29:40 +0100 |
| commit | 98172d46fa866ecaad2afd52777e9ac9a5f15b52 (patch) | |
| tree | 92339227f5cf292d33de7631aca510d40306c97d /candle-examples/examples/segment-anything/model_sam.rs | |
| parent | 28c87f6a34e594aca5f558bceebc4c0a9c95911a (diff) | |
| download | candle-98172d46fa866ecaad2afd52777e9ac9a5f15b52.tar.gz candle-98172d46fa866ecaad2afd52777e9ac9a5f15b52.tar.bz2 candle-98172d46fa866ecaad2afd52777e9ac9a5f15b52.zip | |
Fix some errors about BlockQ8_1 (#776)
* use int8 type instead of uint8 for BlockQ8_1.qs
The uint8 type of BlockQ8_1.qs causes great loss for negative weights
Ref: https://github.com/ggerganov/llama.cpp/blob/ebc96086af49fe70108cafcea6ab4bebd658a41a/ggml.c#L904
Signed-off-by: Zhang Miaolei <zmlcc@outlook.com>
* fix sum error in vec_dot of BlockQ4_1
Ref: https://github.com/ggerganov/llama.cpp/blob/ebc96086af49fe70108cafcea6ab4bebd658a41a/ggml.c#L2840
Signed-off-by: Zhang Miaolei <zmlcc@outlook.com>
* fix sum error in vec_dot of BlockQ5_1
Ref: https://github.com/ggerganov/llama.cpp/blob/ebc96086af49fe70108cafcea6ab4bebd658a41a/ggml.c#L3490
Signed-off-by: Zhang Miaolei <zmlcc@outlook.com>
---------
Signed-off-by: Zhang Miaolei <zmlcc@outlook.com>
Diffstat (limited to 'candle-examples/examples/segment-anything/model_sam.rs')
0 files changed, 0 insertions, 0 deletions