Guard Windows ROCm torchao override skip

studio/install_python_stack.py

@@ -181,6 +181,39 @@ def _probe_installed_torch_version() -> str | None:
     return lines[-1] if lines else None
 
 
+def _installed_torch_is_windows_rocm() -> bool:
+    """Return True when the target venv currently has a Windows ROCm torch build.
+
+    This is a belt-and-suspenders guard for the torchao override step: if the
+    earlier ROCm install path failed to set _rocm_windows_torch_installed but the
+    venv already contains a ROCm torch wheel, still skip torchao because it
+    crashes on import on Windows ROCm.
+    """
+    if not IS_WINDOWS:
+        return False
+    try:
+        probe = subprocess.run(
+            [
+                sys.executable,
+                "-c",
+                (
+                    "import sys, torch; "
+                    "hip = getattr(getattr(torch, 'version', None), 'hip', None) or ''; "
+                    "ver = getattr(torch, '__version__', '').lower(); "
+                    "sys.stdout.write('yes' if (hip or 'rocm' in ver or 'rocmsdk' in ver) else '')"
+                ),
+            ],
+            stdout = subprocess.PIPE,
+            stderr = subprocess.DEVNULL,
+            text = True,
+            timeout = 90,
+            **_windows_hidden_subprocess_kwargs(),
+        )
+    except (OSError, subprocess.TimeoutExpired):
+        return False
+    return probe.returncode == 0 and (probe.stdout or "").strip() == "yes"
+
+
 # AMD Windows ROCm wheels (repo.amd.com/rocm/whl/{arch_family}/).
 # Override with UNSLOTH_ROCM_WINDOWS_MIRROR for air-gapped/mirror installs.
 _ROCM_WINDOWS_INDEX_BASE = (
@@ -2221,7 +2254,7 @@ def _win_amd_smi_has_gpu(stdout: str) -> bool:
     #    (no working build; see below).
     if NO_TORCH:
         _progress("dependency overrides (skipped, no torch)")
-    elif _rocm_windows_torch_installed:
+    elif _rocm_windows_torch_installed or _installed_torch_is_windows_rocm():
         # No working Windows ROCm torchao build: it imports an absent c10d backend
         # and crashes transformers.quantizers. Studio stubs it at runtime, so
         # installing it only ships a package that crashes on import -- skip it.

tests/studio/install/test_rocm_support.py

@@ -2294,6 +2294,41 @@ def _bad_probe(*a, **kw):
         mock_bnb.assert_not_called()
 
 
+class TestWindowsRocmTorchaoGuard:
+    """Verify the torchao skip can detect an installed Windows ROCm torch build."""
+
+    def test_installed_torch_is_windows_rocm_accepts_rocm_probe(self):
+        rv = MagicMock()
+        rv.returncode = 0
+        rv.stdout = "yes"
+        with (
+            patch.object(stack_mod, "IS_WINDOWS", True),
+            patch.object(stack_mod.subprocess, "run", return_value = rv),
+        ):
+            assert stack_mod._installed_torch_is_windows_rocm() is True
+
+    def test_installed_torch_is_windows_rocm_rejects_non_rocm_probe(self):
+        rv = MagicMock()
+        rv.returncode = 0
+        rv.stdout = ""
+        with (
+            patch.object(stack_mod, "IS_WINDOWS", True),
+            patch.object(stack_mod.subprocess, "run", return_value = rv),
+        ):
+            assert stack_mod._installed_torch_is_windows_rocm() is False
+
+    def test_installed_torch_is_windows_rocm_is_non_windows_noop(self):
+        with patch.object(stack_mod, "IS_WINDOWS", False):
+            assert stack_mod._installed_torch_is_windows_rocm() is False
+
+    def test_install_python_stack_uses_direct_windows_rocm_torchao_guard(self):
+        source = _STACK_PATH.read_text(encoding = "utf-8")
+        assert (
+            "elif _rocm_windows_torch_installed or _installed_torch_is_windows_rocm():"
+            in source
+        )
+
+
 # TEST: worker.py -- Windows ROCm patches (source-level checks)
 
 

unsloth/models/rl.py

@@ -1015,6 +1015,9 @@ def _patch_trl_rl_trainers_impl(trainer_file = "grpo_trainer"):
             "dtype = _get_dtype(dtype)\n"
             "float16 = dtype == torch.float16\n"
             "bfloat16 = dtype == torch.bfloat16\n"
+            "if full_finetuning:\n"
+            "    if bfloat16 and use_fp16: use_fp16 = False\n"
+            "    if float16 and use_bf16: use_bf16 = False\n"
             "if not force_float32 and (float16 and use_bf16): raise TypeError('Unsloth: Model is in float16 precision but you want to use bfloat16 precision. Set fp16 to `True` and bf16 to `False`')\n"
             "if not force_float32 and (bfloat16 and use_fp16): raise TypeError('Unsloth: Model is in bfloat16 precision but you want to use float16 precision. Set fp16 to `False` and bf16 to `True`')\n"
             "if force_float32:\n"