From facb8b56f8fd3bb10a693bf0943ae9d69d0828ef Mon Sep 17 00:00:00 2001
From: "Zheng.Deng" <32841220+dengzheng-cloud@users.noreply.github.com>
Date: Wed, 17 Apr 2024 04:51:07 +0800
Subject: [PATCH] convert : fix autoawq gemma (#6704)

* 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.

* change code to full string match and print necessary message

change code to full string match and print a short message to inform users that lm_head.weight has been skipped.

---------

Co-authored-by: Zheng.Deng <32841220+CUGfred@users.noreply.github.com>
---
 convert-hf-to-gguf.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py
index f321d77de..c14186abb 100755
--- a/convert-hf-to-gguf.py
+++ b/convert-hf-to-gguf.py
@@ -2458,6 +2458,12 @@ 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 name == "lm_head.weight":
+                print(f"Skipping get tensor {name!r} in safetensors so that convert can end normally.")
+                continue
+
             old_dtype = data_torch.dtype
 
             # convert any unsupported data types to float32