#!/bin/sh -
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
MODEL_DIR="$(dirname "$SCRIPT_DIR")"
. ~/venv/vllm/bin/activate

dockless destroy seeker-inference
sleep 5
dockless run -d --name seeker-inference \
  -e CUDA_HOME=/usr/local/cuda-13.0 \
  -e C_INCLUDE_PATH=/usr/local/cuda-13.0/include \
  -e LIBRARY_PATH=/usr/lib/aarch64-linux-gnu/nvidia \
  -e FLASHINFER_NVCC=/usr/local/cuda-13.0/bin/nvcc \
  -e VLLM_USE_FLASHINFER_MOE_FP4=0  "$SCRIPT_DIR/unglitched_vllm" \
  --served-model-name Nikola \
  --port 9000 \
  --enable-auto-tool-choice \
  --tool-call-parser minimax_m2 \
  --reasoning-parser minimax_m2_optthink \
  --reasoning-parser-plugin "$SCRIPT_DIR/minimax_m2_optthink_reasoning_parser.py" \
  --enable-prefix-caching \
  --max-num-seqs 4 --cudagraph-capture-sizes 1 2 4 --max-model-len auto \
  --max_num_batched_tokens 8192 \
  --gpu-memory-utilization 0.95 \
  --attention-backend FLASHINFER \
  --async-scheduling \
  --enable-chunked-prefill \
  --chat-template "$SCRIPT_DIR/chat_template.jinja" \
  --model "${@:-$MODEL_DIR}"