router_train_eval_e2e.sh

set -eox pipefail

target_folder="${1}"

# reorient data
./venv/bin/python3.12 reformat_csv.py --input-file "${target_folder}/routing_dataset_merged.csv" --output-file "${target_folder}/routing_data_reformatted.csv"

# split data
./venv/bin/python3.12 split_data.py --input-file "${target_folder}/routing_data_reformatted.csv" --output-train-file "${target_folder}/train.csv" --output-test-file "${target_folder}/test.csv" --train-pct 0.8

# create queries file
./venv/bin/python3.12 create_queries.py --input-train-file "${target_folder}/train.csv" --input-test-file "${target_folder}/test.csv" --output-file "${target_folder}/queries.csv"

# embed queries
./venv/bin/python3.12 ./utils/bert_save_embeddings.py --llm-profiles-file "./router_arena_data/llm_profiles.csv" --queries-file "${target_folder}/queries.csv" --output-dir "${target_folder}/bert_embeddings" --batch-size 16 --max-workers 1

# get cold embeddings
./venv/bin/python3.12 ./utils/cold_test.py --query-embeddings-file "${target_folder}/bert_embeddings/query_embeddings.pkl" --queries-file "${target_folder}/queries.csv" --output-file "${target_folder}/cold_embeddings.pkl"

# train router
bash ./train_router.sh "${target_folder}"

# evaluate the router
bash ./run_a_sweep.sh "${target_folder}" 2>&1 | tee "${target_folder}/sweep.log"

./venv/bin/python3.12 parse_sweep.py --input-file "${target_folder}/sweep.log" --output-file "${target_folder}/parsed_sweep.csv"

./venv/bin/python3.12 plot_comparison.py --baseline-frontier-file "${target_folder}/frontiers/baseline_frontier.json" --pareto-frontier-file "${target_folder}/frontiers/pareto_frontier.json" \
--routing-sweep-file "${target_folder}/parsed_sweep.csv" --output-file "${target_folder}"