diff --git a/pyproject.toml b/pyproject.toml
index f68a194c83..9e489fc320 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -345,8 +345,8 @@ cu130onlytorch2100 = [
     "xformers @ https://download.pytorch.org/whl/cu130/xformers-0.0.34-cp39-abi3-win_amd64.whl ; (sys_platform == 'win32') and (platform_machine == 'AMD64' or platform_machine == 'x86_64')",
 ]
 cu126onlytorch2110 = [
-    # xformers 0.0.35 does not pin torch; pin the trio to +cu126 so it resolves
-    # from the cu126 index (torch 2.11 defaults to a CUDA-13 PyPI wheel).
+    # Pin trio to +cu126 so it resolves from the cu126 index (torch 2.11 defaults
+    # to a CUDA-13 wheel; xformers 0.0.35 does not pin torch).
     "torch==2.11.0+cu126",
     "torchvision==0.26.0+cu126",
     "torchaudio==2.11.0+cu126",
@@ -363,7 +363,7 @@ cu128onlytorch2110 = [
 ]
 cu130onlytorch2110 = [
     # Same +cuNNN pin as cu126/cu128: a bare range lets +cu126 win (PEP 440) and
-    # ===2.11.0 would force-replace cu130-index installs. _auto_install.py adds the index.
+    # === would force-replace cu130-index installs. _auto_install.py adds the index.
     "torch==2.11.0+cu130",
     "torchvision==0.26.0+cu130",
     "torchaudio==2.11.0+cu130",
diff --git a/studio/backend/tests/test_gguf_load_cache_reuse.py b/studio/backend/tests/test_gguf_load_cache_reuse.py
index 62596fcc8a..c70e579092 100644
--- a/studio/backend/tests/test_gguf_load_cache_reuse.py
+++ b/studio/backend/tests/test_gguf_load_cache_reuse.py
@@ -728,11 +728,9 @@ class TestLoadHubDownloadExclusion:
         source = (Path(__file__).resolve().parent.parent / "routes" / "inference.py").read_text()
         gguf_branch = source[source.index("if config.is_gguf:") :]
 
-        # The gguf_load_in_flight marker must be entered before the hub-download
-        # guard and the unload so a concurrent load can't race the download
-        # manager. The llama_extra_args inheritance that used to sit between the
-        # marker and the guard now runs in _guard_chat_load_against_training, ahead
-        # of the GGUF branch, so it is no longer a landmark inside this slice.
+        # gguf_load_in_flight must be entered before the hub-download guard and
+        # unload so a concurrent load can't race the download manager. The old
+        # llama_extra_args landmark moved to _guard_chat_load_against_training.
         assert (
             gguf_branch.index("enter_context(gguf_load_in_flight")
             < gguf_branch.index("_hub_download_blocks_gguf_load")