From 20823530201f96faf4582c1782eb40dcd22fd827 Mon Sep 17 00:00:00 2001
From: "Zheng.Deng" <32841220+CUGfred@users.noreply.github.com>
Date: Mon, 15 Apr 2024 23:44:00 +0800
Subject: [PATCH] fix autoawq quantized gemma model convert error

using autoawq to quantize gemma model will include a lm_head.weight tensor in model-00001-of-00002.safetensors. it result in this situation that convert-hf-to-gguf.py can't map lm_head.weight. skip loading this tensor could prevent this error.
---
 convert-hf-to-gguf.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py
index b51d68307..2e518e71d 100755
--- a/convert-hf-to-gguf.py
+++ b/convert-hf-to-gguf.py
@@ -2262,6 +2262,11 @@ class GemmaModel(Model):
         tensor_map = gguf.get_tensor_name_map(self.model_arch, block_count)
 
         for name, data_torch in self.get_tensors():
+            # lm_head is not used in llama.cpp, while autoawq will include this tensor in model
+            # To prevent errors, skip loading lm_head.weight.
+            if "lm_head.weight" in name:
+                    continue
+
             old_dtype = data_torch.dtype
 
             # convert any unsupported data types to float32