#!/bin/bash
#grab container image name
MODEL_NAME="GLM-5.1-IQ3_KW"
CONT_NAME="ik-$MODEL_NAME"


MODEL="GLM-5.1-IQ3_KS-00001-of-00008.gguf"
MODEL_FOLDER="/sam4t/ENC/LLM/GLM-5.1"

source ./env/ik_env.sh
echo $IMAGE
docker container create --name $CONT_NAME --network llms -p 10987:10987 -v $MODEL_FOLDER:/model --user $(id -u):$(id -g) --gpus=all --restart on-failure:3 \
	--health-cmd "curl -f http://$CONT_NAME:10987/v1/models || exit 1" \
	--health-interval 5s \
	--health-timeout 2s \
	--health-start-period 60s \
	$IMAGE \
	/ik_llama.cpp/build/bin/llama-server \
	--model /model/$MODEL \
	--alias $MODEL_NAME \
	--ctx-size 64000 \
	-amb 512 \
	-muge \
	--merge-qkv \
	-ngl 999 \
	--n-cpu-moe 66 \
	--parallel 1 --threads 32 \
	--host $CONT_NAME \
	--port 10987 --jinja --no-mmap