From 225ec48fe579337efb8fd931e19dc5c6f6078ae4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sigbj=C3=B8rn=20Skj=C3=A6ret?= <sigbjorn.skjaeret@scala.com>
Date: Fri, 14 Jun 2024 13:32:48 +0200
Subject: [PATCH] np.int16 no longer used

---
 convert-hf-to-gguf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py
index ac6363a48..6a3fe2867 100755
--- a/convert-hf-to-gguf.py
+++ b/convert-hf-to-gguf.py
@@ -296,7 +296,7 @@ class Model:
                 if self.ftype != gguf.LlamaFileType.ALL_F32 and extra_f16 and not extra_f32:
                     if self.ftype == gguf.LlamaFileType.MOSTLY_BF16:
                         data = gguf.truncate_bf16(data) if old_dtype == torch.bfloat16 else gguf.quantize_bf16(data)
-                        assert data.dtype in (np.int16, np.uint16)
+                        assert data.dtype == np.uint16
                         data_qtype = gguf.GGMLQuantizationType.BF16
 
                     elif self.ftype == gguf.LlamaFileType.MOSTLY_Q8_0 and gguf.can_quantize_to_q8_0(data):