ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

[pre-commit.ci] auto fixes from pre-commit.com hooks

Browse source 
for more information, see https://pre-commit.ci
This commit is contained in:
pre-commit-ci[bot] 2026-06-01 16:32:28 +00:00
commit e9f52c2114
3 changed files with 36 additions and 24 deletions
Download patch file Download diff file Expand all files Collapse all files

4
studio/backend/tests/test_llama_cpp_context_fit.py
View file

@ -652,7 +652,5 @@ class TestClassifyGpuOffload:
assert inst._classify_gpu_offload(True, [(0, 22805)]) is False
def test_hip_model_buffer_is_gpu(self):
inst = self._backend(
["load_tensors: HIP0 model buffer size = 21000.0 MiB"]
)
inst = self._backend(["load_tensors: HIP0 model buffer size = 21000.0 MiB"])
assert inst._classify_gpu_offload(True, [(0, 22805)]) is True

7
studio/install_llama_prebuilt.py
View file

@ -5919,9 +5919,7 @@ def validate_server(
# the next bundle / source build instead of stopping here.
if _enable_gpu_layers:
log_handle.flush()
offload = server_log_shows_gpu_offload(
read_full_log(log_path)
)
offload = server_log_shows_gpu_offload(read_full_log(log_path))
if offload is False:
raise GpuOffloadFailure(
"llama-server served a completion but loaded the "
@ -5943,8 +5941,7 @@ def validate_server(
raise PrebuiltFallback(
"llama-server served a completion but its startup "
"log carried no GPU-offload signal; smoke-test "
"result is inconclusive:\n"
+ read_log_excerpt(log_path)
"result is inconclusive:\n" + read_log_excerpt(log_path)
)
return
finally:

49
tests/studio/install/test_validate_server_gpu_offload.py
View file

@ -534,7 +534,9 @@ def _gpu_plan(install_kind = "linux-cuda"):
release_tag = "rel",
attempts = [choice],
approved_checksums = M.ApprovedReleaseChecksums(
repo = "unslothai/llama.cpp", release_tag = "rel", upstream_tag = "b9001",
repo = "unslothai/llama.cpp",
release_tag = "rel",
upstream_tag = "b9001",
artifacts = {},
),
)
@ -551,9 +553,12 @@ def _with_server(tmp_path):
def test_existing_cpu_kind_install_is_kept(tmp_path):
# A non-GPU existing install is never offload-gated.
probe = _with_server(tmp_path)
assert M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cpu"), probe
) is True
assert (
M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cpu"), probe
)
is True
)
def test_existing_gpu_install_cpu_only_triggers_reinstall(monkeypatch, tmp_path):
@ -563,17 +568,23 @@ def test_existing_gpu_install_cpu_only_triggers_reinstall(monkeypatch, tmp_path)
raise M.GpuOffloadFailure("loaded the model entirely on CPU")
monkeypatch.setattr(M, "validate_server", fake_validate)
assert M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
) is False
assert (
M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
)
is False
)
def test_existing_gpu_install_offloading_is_kept(monkeypatch, tmp_path):
probe = _with_server(tmp_path)
monkeypatch.setattr(M, "validate_server", lambda *a, **k: None)
assert M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
) is True
assert (
M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
)
is True
)
def test_existing_gpu_install_inconclusive_is_kept(monkeypatch, tmp_path):
@ -583,15 +594,21 @@ def test_existing_gpu_install_inconclusive_is_kept(monkeypatch, tmp_path):
raise PrebuiltFallback("llama-server exited during startup")
monkeypatch.setattr(M, "validate_server", fake_validate)
assert M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
) is True
assert (
M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
)
is True
)
def test_existing_gpu_install_no_binary_is_kept(tmp_path):
# No llama-server present -> let the normal flow reinstall, don't crash.
probe = tmp_path / "probe.gguf"
probe.write_bytes(b"GGUF")
assert M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
) is True
assert (
M.existing_gpu_install_offloads(
tmp_path, nvidia_host(), _gpu_plan("linux-cuda"), probe
)
is True
)