[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
@@ -20,12 +20,8 @@ __all__ = ["get_parameter_for_reload", "reload_weights"]
def get_parameter_for_reload(model: nn.Module, name: str) -> nn.Parameter:
"""Resolve checkpoint names without changing the model's module tree."""
from vllm.lora.layers import BaseLayerWithLoRA
module_name, _, parameter_name = name.rpartition(".")
module = model.get_submodule(module_name)
if isinstance(module, BaseLayerWithLoRA):
module = module.base_layer
return module.get_parameter(parameter_name)
@@ -39,7 +35,7 @@ def reload_weights(
"""Reload a model's weights in place from a weights iterator or from disk.
Callers are responsible for invalidating any state derived from the old
weights (e.g. LoRA, encoder and multimodal caches).
weights (e.g. encoder and multimodal caches).
Args:
vllm_config: config of the model being reloaded. If `weights_path` is
@@ -63,10 +59,7 @@ def reload_weights(
model_config = vllm_config.model_config
load_config = vllm_config.load_config
weights_to_load = {
name.replace(".base_layer.", ".") if vllm_config.lora_config else name
for name, _ in model.named_parameters()
}
weights_to_load = {name for name, _ in model.named_parameters()}
counter_before_reloading = time.perf_counter()
# load weights from disk if none are provided