llama : fix comment for "output.weight" tensor

1 year ago · d40fded93e
parent 2510c1831f
commit d40fded93e
1 changed files with 2 additions and 2 deletions
--- a/llama.cpp
+++ b/llama.cpp
@ -1618,8 +1618,8 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
        // quantize only 2D tensors
        quantize &= (tensor.ne.size() == 2);

-        // GG: uncomment this to keep the output layer in FP16
-        //if (tensor.name.rfind("output")) {
+        // uncomment this to keep the output layer in FP16
+        //if (tensor.name == "output.weight") {
        //    quantize = false;
        //}