diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2024-12-18 13:29:25 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-12-18 13:29:25 +0100 |
| commit | 59d742b00fb48a3704f86c16afbb6b4ebcde8e68 (patch) | |
| tree | ed95493397cc9ea45b8f8d897d1200d0b3588a86 /include | |
| parent | 9b6d14a2991da41af4aa7ef64a712c63b73ad9fe (diff) | |
IQ5_K_R4 (#149)
* iq5_k_r4: Zen4
Much slower than the others.
* iq5_k_r5: WIP
* Minor
* iq5_k_r4: fix AVX2 nrc_y = 1 case
* iq5_k_r4: better Zen4
But TG is still slower than iq5_k
* iq5_k_r4: slightly better AVX2
* iq5_k_r4: NEON
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'include')
| -rw-r--r-- | include/llama.h | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/include/llama.h b/include/llama.h index e63d76fe..7267fbd4 100644 --- a/include/llama.h +++ b/include/llama.h @@ -196,6 +196,7 @@ extern "C" { LLAMA_FTYPE_MOSTLY_IQ2_K_R4 = 338, // except 1d tensors LLAMA_FTYPE_MOSTLY_IQ3_K_R4 = 339, // except 1d tensors LLAMA_FTYPE_MOSTLY_IQ4_K_R4 = 340, // except 1d tensors + LLAMA_FTYPE_MOSTLY_IQ5_K_R4 = 341, // except 1d tensors LLAMA_FTYPE_MOSTLY_Q8_K_R8 = 399, // except 1d tensors LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file |