run_inference_loop.sh

#!/usr/bin/env bash

PRETRAINED=${1:-"hpcaitech/grok-1"}
TOKENIZER=${2:-"tokenizer.model"}

while true; do
    echo "Enter the text you want to process:"
    read USER_TEXT
    
    torchrun --standalone --nproc_per_node 8 inference_tp.py --pretrained "$PRETRAINED" \
        --tokenizer "$TOKENIZER" \
        --max_new_tokens 64 \
        --text "$USER_TEXT"
done