diff options
author | Georgi Gerganov <ggerganov@gmail.com> | 2023-04-21 10:23:36 +0300 |
---|---|---|
committer | Georgi Gerganov <ggerganov@gmail.com> | 2023-04-21 10:24:02 +0300 |
commit | d40fded93e1a533e969768e1e335c15c61c296ce (patch) | |
tree | f2a3e8e145196a1521f3031f7165d92a7069b8fc | |
parent | 2510c1831fac874f32e272f6079f01b5461f3986 (diff) |
llama : fix comment for "output.weight" tensor
-rw-r--r-- | llama.cpp | 4 |
1 files changed, 2 insertions, 2 deletions
@@ -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; //} |