[Mini-VLLM] Remove LoRA, plugins, speech-to-text and scale-out surfaces

Production runs one static model behind the OpenAI-compatible API. Every
adapter path and non-OpenAI HTTP surface is unreachable there.

- Delete the LoRA stack: vllm/lora, LoRARequest on the engine wire, the
  scheduler, worker, cudagraph and MoE plumbing, LoRA metrics, the lora
  CLI group, and the LoRA tests.
- Drop the plugin loader in vllm/plugins. Platform resolution is now
  CUDA-only and the general/stat-logger plugin hooks are gone.
- Delete the speech-to-text and realtime entrypoints, the scale-out
  render routes, and vllm/config/speech_to_text.py. GenerationTask is
  generate-only.
- Cut DecodeBenchConnector and the disagg-prefill doc and image. The
  Multi, Offloading and Example connectors stay; production uses the
  CPU KV offload tier.
- Delete the ROCm-only fused KDA chunk kernel (2346 lines) with its CMake
  arch gate, declarations and op registrations. The CUDA build never
  defined the guard and no Python caller existed.
- Delete the orphan Mistral and Pixtral chat fixtures and examples/.
- Fix _check_model: removing the LoRA early return made every request
  fail with "The model ... does not exist. Valid aliases: ...".

Verified on an RTX 3090: import sweep reports 1242 modules with 7 known
optional-dependency failures; a live server answers chat, completions,
responses, tokenize and detokenize with zero log errors and 404 for the
removed surfaces; collection yields 8558 tests with the 3 pre-existing
errors; ruff adds no new finding; the incremental _C build succeeds.

Co-authored-by: Hinata <hinata@idihore.id>
Signed-off-by: MOVZX <movzx@yahoo.com>
This commit is contained in:
2026-10-19 21:06:19 +07:00
co-authored by Hinata
parent 5f3c7bd8ab
commit 5aa2ac64f8
269 changed files with 417 additions and 29932 deletions
@@ -592,13 +592,12 @@ class TestLoadWeightsExpertMapping:
return calls
@pytest.mark.parametrize(
"projs,lora_prefix",
"projs",
[
(("gate_proj", "down_proj", "up_proj"), ""),
(("w1", "w2", "w3"), ""),
(("up_proj", "down_proj", "up_proj"), ""),
(("up_proj", "down_proj", None), ""),
(("gate_proj", "down_proj", "up_proj"), "base_layer."),
("gate_proj", "down_proj", "up_proj"),
("w1", "w2", "w3"),
("up_proj", "down_proj", "up_proj"),
("up_proj", "down_proj", None),
],
)
@pytest.mark.parametrize(
@@ -610,12 +609,11 @@ class TestLoadWeightsExpertMapping:
],
)
def test_per_expert_weights_keep_all_physical_destinations(
self, projs, lora_prefix, suffix, shape, dtype
self, projs, suffix, shape, dtype
):
mapping = self._mapping(
projs,
num_redundant_experts=4,
lora_base_layer_prefix=lora_prefix,
is_gated=projs[2] is not None,
)
for logical_id, physical_ids in [(1, (1, 17)), (10, (10,))]:
@@ -623,7 +621,7 @@ class TestLoadWeightsExpertMapping:
if not proj:
continue
weight = torch.full(shape, logical_id + 1, dtype=dtype)
name = f"{logical_id}.{proj}.{lora_prefix}{suffix}"
name = f"{logical_id}.{proj}.{suffix}"
calls = self._load(mapping, [(name, weight)], suffix)
shards = [s for p, s in zip(projs, ("w1", "w2", "w3")) if p == proj]
expected = [
@@ -636,20 +634,18 @@ class TestLoadWeightsExpertMapping:
torch.testing.assert_close(call[3], weight, rtol=0, atol=0)
@pytest.mark.parametrize(
"projs,fused_name,lora_prefix",
"projs,fused_name",
[
(("gate_proj", "down_proj", "up_proj"), "gate_up_proj", ""),
(("w1", "w2", "w3"), "w13", "base_layer."),
(("gate_proj", "down_proj", "up_proj"), "gate_up_proj"),
(("w1", "w2", "w3"), "w13"),
],
)
def test_per_expert_fused_gate_up_keeps_both_halves(
self, projs, fused_name, lora_prefix
self, projs, fused_name
):
mapping = self._mapping(
projs, num_redundant_experts=4, lora_base_layer_prefix=lora_prefix
)
mapping = self._mapping(projs, num_redundant_experts=4)
weight = torch.arange(12).reshape(4, 3)
calls = self._load(mapping, [(f"1.{fused_name}.{lora_prefix}weight", weight)])
calls = self._load(mapping, [(f"1.{fused_name}.weight", weight)])
assert [call[:3] for call in calls] == [
("w13_weight", shard, expert)
for expert in (1, 17)