From 7f5b7b368cb4d4c509ce8a88ce2d80eb2937cd9a Mon Sep 17 00:00:00 2001
From: "pre-commit-ci[bot]"
 <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Tue, 17 Mar 2026 08:24:17 +0000
Subject: [PATCH] [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci
---
 unsloth/kernels/fp8.py   |  5 ++---
 unsloth/models/loader.py | 16 ++++++++++------
 2 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/unsloth/kernels/fp8.py b/unsloth/kernels/fp8.py
index e3fd1b9cc4..044e5a9292 100644
--- a/unsloth/kernels/fp8.py
+++ b/unsloth/kernels/fp8.py
@@ -605,9 +605,8 @@ except:
     pass
 
 
-HAS_FBGEMM_FP8_OPS = (
-    hasattr(torch.ops, "fbgemm")
-    and hasattr(torch.ops.fbgemm, "quantize_fp8_per_row")
+HAS_FBGEMM_FP8_OPS = hasattr(torch.ops, "fbgemm") and hasattr(
+    torch.ops.fbgemm, "quantize_fp8_per_row"
 )
 
 
diff --git a/unsloth/models/loader.py b/unsloth/models/loader.py
index 43092bc71d..38811257ff 100644
--- a/unsloth/models/loader.py
+++ b/unsloth/models/loader.py
@@ -819,7 +819,9 @@ class FastLanguageModel(FastLlamaModel):
 
         if load_in_fp8 != False:
             _tag_model_with_fp8_torchao_config(model, fp8_mode)
-        if load_in_fp8 != False or _has_prequantized_fp8_config(model_name, token=token, trust_remote_code=trust_remote_code):
+        if load_in_fp8 != False or _has_prequantized_fp8_config(
+            model_name, token = token, trust_remote_code = trust_remote_code
+        ):
             maybe_patch_stacked_moe_expert_fp8_scales(
                 model,
                 model_name = model_name,
@@ -1363,10 +1365,10 @@ class FastModel(FastBaseModel):
                 model_name = get_model_name(
                     model_name,
                     load_in_4bit,
-                    load_in_fp8=load_in_fp8,
-                    fast_inference=fast_inference,
-                    token=token,
-                    trust_remote_code=trust_remote_code,
+                    load_in_fp8 = load_in_fp8,
+                    fast_inference = fast_inference,
+                    token = token,
+                    trust_remote_code = trust_remote_code,
                 )
             # Check if pre-quantized models are allowed
             # AMD Instinct GPUs need blocksize = 128 on bitsandbytes < 0.49.2 (our pre-quants use blocksize = 64)
@@ -1576,7 +1578,9 @@ class FastModel(FastBaseModel):
 
         if load_in_fp8 != False:
             _tag_model_with_fp8_torchao_config(model, fp8_mode)
-        if load_in_fp8 != False or _has_prequantized_fp8_config(model_name, token=token, trust_remote_code=trust_remote_code):
+        if load_in_fp8 != False or _has_prequantized_fp8_config(
+            model_name, token = token, trust_remote_code = trust_remote_code
+        ):
             maybe_patch_stacked_moe_expert_fp8_scales(
                 model,
                 model_name = model_name,