Merge main into dev/vllm-align for vLLM 0.31 release

Signed-off-by: tzhouam <tzhouam@connect.ust.hk>
This commit is contained in:
tzhouam committed 2026-10-03 21:05:17 +08:00
commit 008fa25263
252 files changed
+27589 -1237

No files matched your search

+3
View File
@@ -198,6 +198,7 @@ class OmniEngineArgs(EngineArgs):
async_chunk: bool = False
session_mode: str = "turn"
retains_state_across_chunks: bool = False
supports_running_prefix_cache_reset: bool = True
use_v2_model_runner: bool = False
supports_native_mrv2_data_plane: bool = False
# WS-A: Stage-1 active stream slots. 0 = legacy preempt-everything.
@@ -432,6 +433,7 @@ class OmniEngineArgs(EngineArgs):
async_chunk=self.async_chunk,
session_mode=self.session_mode,
retains_state_across_chunks=self.retains_state_across_chunks,
supports_running_prefix_cache_reset=self.supports_running_prefix_cache_reset,
use_v2_model_runner=self.use_v2_model_runner,
supports_native_mrv2_data_plane=self.supports_native_mrv2_data_plane,
active_stream_window=self.active_stream_window,
@@ -560,6 +562,7 @@ class OrchestratorArgs:
# === Diffusion model config ===
num_gpus: int | None = None
model_class_name: str | None = None
hsdp_weight_load_strategy: str | None = None
diffusion_load_format: str | None = None
lora_path: list[str] | None = None
lora_backend: str | None = None