diff options
| author | Nexes the Elder <124105151+Nexesenex@users.noreply.github.com> | 2024-11-21 07:12:57 +0100 |
|---|---|---|
| committer | Iwan Kawrakow <iwan.kawrakow@gmail.com> | 2024-11-21 12:01:23 +0200 |
| commit | 8ad84b9fab9570c36220cb791f9a67a4d2c7fd2f (patch) | |
| tree | 6fd6144c0abdb737298775784273a1fb246713ea /src | |
| parent | 4d2fbde0cbbfc98200b59ed6fe5b32628a70c055 (diff) | |
Use Q6_0 instead of Q5_1 for tensors incompatible with IQ5_K/Q5_K (#116)
Diffstat (limited to 'src')
| -rw-r--r-- | src/llama.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/llama.cpp b/src/llama.cpp index 2b9a1b1a..61448319 100644 --- a/src/llama.cpp +++ b/src/llama.cpp @@ -16048,7 +16048,7 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n case GGML_TYPE_IQ4_K: case GGML_TYPE_Q4_K: new_type = GGML_TYPE_Q5_0; break; case GGML_TYPE_IQ5_K: - case GGML_TYPE_Q5_K: new_type = GGML_TYPE_Q5_1; break; + case GGML_TYPE_Q5_K: new_type = GGML_TYPE_Q6_0; break; case GGML_TYPE_IQ6_K: case GGML_TYPE_Q6_K: new_type = GGML_TYPE_Q8_0; break; default: throw std::runtime_error("\nUnsupported tensor size encountered\n"); |