Inconsistent time logging for speech_recognition benchmark

[jbalma]

Currently the directions and scripts for the speech_recognition benchmark assume we time the execution of the run_and_time.sh script like this:

cd pytorch
cd docker
sh build-docker.sh
sh run-dev.sh
time sh run_and_time.sh | tee speech_ds2.out

However, the other benchmarks use the date command (instead of time) from within the run_and_time.sh script itself. For example, the timing for the translation benchmark's run_and_time.sh script contains this timing block:

---

Start timing

START=$(date +%s)
START_FMT=$(date +%Y-%m-%d\ %r)
echo "STARTING TIMING RUN AT ${START_FMT}"

Run benchmark (training)

SEED=${1:-1}

echo "Running benchmark with seed ${SEED}"
. run_training.sh ${SEED} ${TARGET_UNCASED_BLEU_SCORE}

RET_CODE=$?; if [[ ${RET_CODE} != 0 ]]; then exit ${RET_CODE}; fi

End timing

END=$(date +%s)
END_FMT=$(date +%Y-%m-%d\ %r)

echo "ENDING TIMING RUN AT ${END_FMT}"

Report result

result=$(( ${END} - ${START} ))
result_name="transformer"

We should update the speech_recognition benchmark's run_and_time.sh script to match the others for consistency , but it's probably not critical for the 0.5 release.

-Jacob