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

Update llama.cpp - exception for the IQ2_S token embedding error

Browse source
This commit is contained in:
Nexesenex 2024-03-26 10:11:46 +01:00 • committed by GitHub
parent d1839362fc
commit eaf9571d9b
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
llama.cpp
View file

@ -12984,8 +12984,8 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
} }
if ((new_type == GGML_TYPE_IQ2_XXS || if ((new_type == GGML_TYPE_IQ2_XXS ||
new_type == GGML_TYPE_IQ2_XS || new_type == GGML_TYPE_IQ2_XS ||
new_type == GGML_TYPE_IQ2_S ||
new_type == GGML_TYPE_IQ1_S || new_type == GGML_TYPE_IQ1_S ||
(new_type == GGML_TYPE_IQ2_S && strcmp(tensor->name, "token_embd.weight")) ||
(new_type == GGML_TYPE_Q2_K && params->ftype == LLAMA_FTYPE_MOSTLY_Q2_K_S && strcmp(tensor->name, "token_embd.weight") != 0)) && !imatrix) { (new_type == GGML_TYPE_Q2_K && params->ftype == LLAMA_FTYPE_MOSTLY_Q2_K_S && strcmp(tensor->name, "token_embd.weight") != 0)) && !imatrix) {
LLAMA_LOG_ERROR("\n\n============================================================\n"); LLAMA_LOG_ERROR("\n\n============================================================\n");
LLAMA_LOG_ERROR("Missing importance matrix for tensor %s in a very low-bit quantization\n", tensor->name); LLAMA_LOG_ERROR("Missing importance matrix for tensor %s in a very low-bit quantization\n", tensor->name);