[Rebase] Rebase to vllm 0.20.0 (#3232)

Signed-off-by: tzhouam <tzhouam@connect.ust.hk>
Signed-off-by: Zhou Taichang <tzhouam@connect.ust.hk>
Signed-off-by: Yuanheng Zhao <jonathan.zhaoyh@gmail.com>
Co-authored-by: Yuanheng Zhao <54058983+yuanheng-zhao@users.noreply.github.com>
This commit is contained in:
Zhou TaichangandYuanheng Zhao authored and GitHub committed 2026-04-30 09:56:59 +12:00
1 parent dd0fa02547
commit 7f2b1e9386
70 files changed
+1191 -273

No files matched your search

@@ -411,10 +411,10 @@ def test_gpu_ar_model_runner_supplies_req_output_history_to_model_sampler():
runner.input_batch = _DummyInputBatch()
runner.model = SimpleNamespace(
prefer_model_sampler=True,
sample=lambda logits, sampling_metadata: seen_histories.append(
[list(x) for x in sampling_metadata.output_token_ids]
)
or SamplerOutput(sampled_token_ids=torch.tensor([[7]], dtype=torch.int32), logprobs_tensors=None),
sample=lambda logits, sampling_metadata: (
seen_histories.append([list(x) for x in sampling_metadata.output_token_ids])
or SamplerOutput(sampled_token_ids=torch.tensor([[7]], dtype=torch.int32), logprobs_tensors=None)
),
)
runner.sampler = lambda **_: (_ for _ in ()).throw(AssertionError("fallback sampler should not be used"))
@@ -450,10 +450,10 @@ def test_gpu_ar_model_runner_repairs_async_placeholders_for_model_sampler():
runner.input_batch = _DummyInputBatch()
runner.model = SimpleNamespace(
prefer_model_sampler=True,
sample=lambda logits, sampling_metadata: seen_histories.append(
[list(x) for x in sampling_metadata.output_token_ids]
)
or SamplerOutput(sampled_token_ids=torch.tensor([[7]], dtype=torch.int32), logprobs_tensors=None),
sample=lambda logits, sampling_metadata: (
seen_histories.append([list(x) for x in sampling_metadata.output_token_ids])
or SamplerOutput(sampled_token_ids=torch.tensor([[7]], dtype=torch.int32), logprobs_tensors=None)
),
)
runner.sampler = lambda **_: (_ for _ in ()).throw(AssertionError("fallback sampler should not be used"))
@@ -0,0 +1,22 @@
from __future__ import annotations
from transformers import PreTrainedModel
from vllm_omni.model_executor.models.dynin_omni.dynin_omni_token2audio import (
_ensure_transformers_tied_weights_compat,
)
def test_transformers_tied_weights_compat(monkeypatch):
monkeypatch.delattr(PreTrainedModel, "all_tied_weights_keys", raising=False)
_ensure_transformers_tied_weights_compat()
model = PreTrainedModel.__new__(PreTrainedModel)
model._tied_weights_keys = ["decoder.weight"]
model._dynamic_tied_weights_keys = {"lm_head.weight": "decoder.weight"}
assert model.all_tied_weights_keys == {
"decoder.weight": None,
"lm_head.weight": "decoder.weight",
}