quantize : fix precedence of cli args (#6541)
This commit is contained in:
parent
e3c337d87c
commit
b73e564b16
1 changed files with 2 additions and 2 deletions
|
@ -13562,10 +13562,10 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
|
|||
if (!params->pure && ggml_is_quantized(default_type)) {
|
||||
new_type = llama_tensor_get_type(qs, new_type, tensor, ftype);
|
||||
}
|
||||
else if (params->token_embedding_type < GGML_TYPE_COUNT && strcmp(tensor->name, "token_embd.weight") == 0) {
|
||||
if (params->token_embedding_type < GGML_TYPE_COUNT && strcmp(tensor->name, "token_embd.weight") == 0) {
|
||||
new_type = params->token_embedding_type;
|
||||
}
|
||||
else if (params->output_tensor_type < GGML_TYPE_COUNT && strcmp(tensor->name, "output.weight") == 0) {
|
||||
if (params->output_tensor_type < GGML_TYPE_COUNT && strcmp(tensor->name, "output.weight") == 0) {
|
||||
new_type = params->output_tensor_type;
|
||||
}
|
||||
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue