#!/usr/bin/env bash set -euo pipefail TASK_ROOT=$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")/.." && pwd) docker run -d --name embeddinggemma2-q4-validation \ --gpus all \ --publish 127.0.0.1:12435:12434 \ --mount type=volume,src=embeddinggemma2-q4-validation-models,dst=/models \ --mount "type=bind,src=${TASK_ROOT}/llama.cpp/build-cuda/bin,dst=/opt/llama,readonly" \ --env MODEL_RUNNER_PORT=12434 \ --env LLAMA_SERVER_PATH=/opt/llama \ --env LD_LIBRARY_PATH=/opt/llama:/app \ --env DISABLE_SERVER_UPDATE=1 \ --env 'LLAMA_ARGS=-ngl 999 --pooling mean -b 2048 -ub 2048 --parallel 1 -t 10' \ docker/model-runner@sha256:9967b6c1092ff01e7cf68119ab25067a1f58e3d1b06873bc16fe1e758bcd914f