#!/bin/bash
#
# Print the model name, max context size, and average tokens per second from a
# llama-server log file:
#
#   llama_server_model_ctx_and_aver_tokens_per_sec.sh <log-file>
#
log_file=$1
script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
grep "general.name *=" "${log_file}"
grep "llama_context: n_ctx " "${log_file}"
grep "  eval time" "${log_file}" | "${script_dir}/llama_server_average_tokens_per_sec.py"