Deprecations for v0.23 and v0.24 (#44992)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2026-06-11 14:35:38 +00:00
committed by GitHub
parent 55911db580
commit 03878d1c22
35 changed files with 100 additions and 674 deletions
@@ -106,7 +106,7 @@ def test_reward_models_using_activation(
dtype=dtype,
pooler_config=PoolerConfig(use_activation=False),
) as vllm_model:
wo_activation = vllm_model.reward(example_prompts)
wo_activation = vllm_model.token_classify(example_prompts)
with vllm_runner(
model,
@@ -114,7 +114,7 @@ def test_reward_models_using_activation(
dtype=dtype,
pooler_config=PoolerConfig(use_activation=True),
) as vllm_model:
w_activation = vllm_model.reward(example_prompts)
w_activation = vllm_model.token_classify(example_prompts)
for wo, w in zip(wo_activation, w_activation):
wo = torch.tensor(wo)
+2 -2
View File
@@ -107,7 +107,7 @@ def test_prm_models(
pytest.skip("CPU only supports V1")
with vllm_runner(model, max_model_len=1024, dtype=dtype) as vllm_model:
vllm_outputs = vllm_model.reward(math_step_prompts)
vllm_outputs = vllm_model.token_classify(math_step_prompts)
with hf_runner(model, dtype=dtype, auto_cls=AutoModel) as hf_model:
hf_model = step_reward_patch_hf_model(hf_model)
@@ -146,7 +146,7 @@ def test_prm_models_with_golden_outputs(
pytest.skip(f"No available golden outputs for {model}.")
with vllm_runner(model, max_model_len=1024, dtype=dtype) as vllm_model:
vllm_outputs = vllm_model.reward(math_step_prompts)
vllm_outputs = vllm_model.token_classify(math_step_prompts)
golden_outputs = load_reward_outputs(FIXTURE_REWARD_RESULT[model])
+2 -2
View File
@@ -133,11 +133,11 @@ def test_nvfp4(vllm_runner, model, eager, backend):
not current_platform.is_rocm(),
reason="NVFP4 MOE emulation is only useful on AMD Instinct MI3xx",
)
def test_nvfp4_moe(vllm_runner, model, backend, monkeypatch):
monkeypatch.setenv("VLLM_NVFP4_GEMM_BACKEND", backend)
def test_nvfp4_moe(vllm_runner, model, backend):
with vllm_runner(
model,
moe_backend=backend,
linear_backend=backend,
load_format="dummy",
hf_overrides={"num_hidden_layers": 2},
) as llm: