lvllm-v2.3.5

feat: update lk_moe==2.2.2 to support dynamic FP8 quantization granularity (groupK/groupN), for example : deepreinforce-ai/Ornith-1.0-35B-FP8

PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True \
LVLLM_MOE_NUMA_ENABLED=1 \
LK_THREADS=44 \
OMP_NUM_THREADS=44 \
LK_THREAD_BINDING=CPU_CORE \
LVLLM_GPU_RESIDENT_MOE_LAYERS=0-4 \
LVLLM_GPU_PREFETCH_WINDOW=1 \
LVLLM_GPU_PREFILL_MIN_BATCH_SIZE=128 \
LK_POWER_SAVING=1 \
vllm serve /home/guqiong/Models/Ornith-1.0-35B-FP8 \
  --host 0.0.0.0 \
  --port 8070 \
  --tensor-parallel-size 2 \
  --max-model-len 128000 \
  --gpu-memory-utilization 0.95 \
  --trust-remote-code \
  --tokenizer-mode auto \
  --served-model-name Ornith-1.0-35B-FP8 \
  --compilation-config '{"cudagraph_mode": "FULL_DECODE_ONLY", "mode": "VLLM_COMPILE"}' \
  --enable-prefix-caching \
  --enable-chunked-prefill \
  --max-num-batched-tokens 36000 \
  --dtype bfloat16 \
  --max-num-seqs 2 \
  --enable-auto-tool-choice \
  --reasoning-parser qwen3 \
  --tool-call-parser qwen3_coder