From 640ff681c388bde8f9348306ce2b90357878bc1c Mon Sep 17 00:00:00 2001 From: chenyu Date: Thu, 3 Apr 2025 23:36:24 -0400 Subject: [PATCH] rename bert script to 8xMI300X (#9734) and adds a script for single MI300X --- .../implementations/tinybox_1xMI300X/dev_beam.sh | 15 +++++++++++++++ .../{tinybox_amd => tinybox_8xMI300X}/dev_beam.sh | 0 .../{tinybox_amd => tinybox_8xMI300X}/dev_run.sh | 0 3 files changed, 15 insertions(+) create mode 100755 examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_1xMI300X/dev_beam.sh rename examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/{tinybox_amd => tinybox_8xMI300X}/dev_beam.sh (100%) rename examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/{tinybox_amd => tinybox_8xMI300X}/dev_run.sh (100%) diff --git a/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_1xMI300X/dev_beam.sh b/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_1xMI300X/dev_beam.sh new file mode 100755 index 0000000000..452fb55722 --- /dev/null +++ b/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_1xMI300X/dev_beam.sh @@ -0,0 +1,15 @@ +#!/bin/bash + +export PYTHONPATH="." +export MODEL="bert" +export DEFAULT_FLOAT="HALF" GPUS=1 BS=128 EVAL_BS=128 + +export BEAM=3 BEAM_UOPS_MAX=4000 BEAM_UPCAST_MAX=256 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 +export IGNORE_JIT_FIRST_BEAM=1 +# export BEAM_LOG_SURPASS_MAX=1 +# export BASEDIR="/raid/datasets/wiki" + +export RESET_STEP=1 +export BENCHMARK=10 DEBUG=2 + +python3 examples/mlperf/model_train.py diff --git a/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_amd/dev_beam.sh b/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/dev_beam.sh similarity index 100% rename from examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_amd/dev_beam.sh rename to examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/dev_beam.sh diff --git a/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_amd/dev_run.sh b/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/dev_run.sh similarity index 100% rename from examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_amd/dev_run.sh rename to examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/dev_run.sh