diff options
| author | Stephan Walter <stephan@walter.name> | 2023-04-28 23:10:43 +0000 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-04-28 23:10:43 +0000 |
| commit | 36d19a603b221d1bd7897fcb10e823e2103b052d (patch) | |
| tree | 14e8caf8b61f8fa0fa72f5d7cc137b1ff9330b2f /llama.h | |
| parent | 7f15c5c477d9933689a9d1c40794483e350c2f19 (diff) | |
Remove Q4_3 which is no better than Q5 (#1218)
Diffstat (limited to 'llama.h')
| -rw-r--r-- | llama.h | 2 |
1 files changed, 1 insertions, 1 deletions
@@ -73,7 +73,7 @@ extern "C" { LLAMA_FTYPE_MOSTLY_Q4_1 = 3, // except 1d tensors LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16 LLAMA_FTYPE_MOSTLY_Q4_2 = 5, // except 1d tensors - LLAMA_FTYPE_MOSTLY_Q4_3 = 6, // except 1d tensors + // LLAMA_FTYPE_MOSTLY_Q4_3 (6) support has been removed LLAMA_FTYPE_MOSTLY_Q8_0 = 7, // except 1d tensors LLAMA_FTYPE_MOSTLY_Q5_0 = 8, // except 1d tensors LLAMA_FTYPE_MOSTLY_Q5_1 = 9, // except 1d tensors |