From: Bartowski <redacted>
Date: Thu, 2 May 2024 23:49:09 +0000 (-0400)
Subject: Remove .attention from skipped tensors to match more accurately (#7051)
X-Git-Tag: upstream/0.0.4488~1705
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=60325fa56f61c228464c9f065db3aa6a61f2156e;p=pkg%2Fggml%2Fsources%2Fllama.cpp

Remove .attention from skipped tensors to match more accurately (#7051)
---

diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py
index 2f146d73..612aea17 100755
--- a/convert-hf-to-gguf.py
+++ b/convert-hf-to-gguf.py
@@ -1427,7 +1427,7 @@ class LlamaModel(Model):
         experts = dict()
         for name, data_torch in self.get_tensors():
             # we don't need these
-            if name.endswith((".attention.masked_bias", ".attention.bias", ".attention.rotary_emb.inv_freq")):
+            if name.endswith((".attention.masked_bias", ".attention.bias", ".rotary_emb.inv_freq")):
                 continue
 
             old_dtype = data_torch.dtype