From 7fa568dee3096475771fc5c977cdfce5c46ffbdb Mon Sep 17 00:00:00 2001
From: "zhao.wang" <57819425+Excelius-Wang@users.noreply.github.com>
Date: Tue, 25 Aug 2026 19:14:24 +0800
Subject: [PATCH] docs: correct Adagrad and matrix multiplication wording

---
 bitsandbytes/functional.py    | 2 +-
 bitsandbytes/optim/adagrad.py | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/bitsandbytes/functional.py b/bitsandbytes/functional.py
index 33b2cd9ac..0b32b1609 100644
--- a/bitsandbytes/functional.py
+++ b/bitsandbytes/functional.py
@@ -1291,7 +1291,7 @@ def check_matmul(A, B, out, transposed_A, transposed_B, expected_type=torch.int8
 
     if not correct:
         raise ValueError(
-            f"Tensor dimensions incorrect for matrix mulitiplication: A x B: {sA} x {sB} with transpose for A x B: {tA} x {tB}.",
+            f"Tensor dimensions incorrect for matrix multiplication: A x B: {sA} x {sB} with transpose for A x B: {tA} x {tB}.",
         )
 
     return sout
diff --git a/bitsandbytes/optim/adagrad.py b/bitsandbytes/optim/adagrad.py
index 2566d1e47..416d64801 100644
--- a/bitsandbytes/optim/adagrad.py
+++ b/bitsandbytes/optim/adagrad.py
@@ -31,7 +31,7 @@ def __init__(
             weight_decay (`float`, defaults to 0.0):
                 The weight decay value for the optimizer.
             initial_accumulator_value (`int`, defaults to 0):
-                The initial momemtum values.
+                The initial accumulator value.
             eps (`float`, defaults to 1e-10):
                 The epsilon value prevents division by zero in the optimizer.
             optim_bits (`int`, defaults to 32):
@@ -90,7 +90,7 @@ def __init__(
             weight_decay (`float`, defaults to 0.0):
                 The weight decay value for the optimizer.
             initial_accumulator_value (`int`, defaults to 0):
-                The initial momemtum values.
+                The initial accumulator value.
             eps (`float`, defaults to 1e-10):
                 The epsilon value prevents division by zero in the optimizer.
             optim_bits (`int`, defaults to 8):
@@ -154,7 +154,7 @@ def __init__(
             weight_decay (`float`, defaults to 0.0):
                 The weight decay value for the optimizer.
             initial_accumulator_value (`int`, defaults to 0):
-                The initial momemtum values.
+                The initial accumulator value.
             eps (`float`, defaults to 1e-10):
                 The epsilon value prevents division by zero in the optimizer.
             optim_bits (`int`, defaults to 32):