From 2787f598b85f327b2920c45916b8c6d62998ac6a Mon Sep 17 00:00:00 2001
From: Rohan170603 <163886014+Rohan170603@users.noreply.github.com>
Date: Mon, 10 Aug 2026 21:47:28 +0530
Subject: [PATCH] docs: fix float16 accumulation data type

---
 docs/source/concept_guides/quantization.mdx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/source/concept_guides/quantization.mdx b/docs/source/concept_guides/quantization.mdx
index e7d934ba98..688fba5d41 100644
--- a/docs/source/concept_guides/quantization.mdx
+++ b/docs/source/concept_guides/quantization.mdx
@@ -15,7 +15,7 @@ The basic idea behind quantization is quite easy: going from high-precision repr
 floating-point) for weights and activations to a lower precision data type. The most common lower precision data types
 are:
 
-- `float16`, accumulation data type `float16`
+- `float16`, accumulation data type `float32`
 - `bfloat16`, accumulation data type `float32`
 - `int16`, accumulation data type `int32`
 - `int8`, accumulation data type `int32`