From eeb5a28fa2ef6f44bfd03ab3a29ffb33d7fafa51 Mon Sep 17 00:00:00 2001 From: Arjun Suresh Date: Wed, 12 Feb 2025 22:18:29 +0000 Subject: [PATCH 1/8] Update test-nvidia-mlperf-inference-implementations.yml --- .../workflows/test-nvidia-mlperf-inference-implementations.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/test-nvidia-mlperf-inference-implementations.yml b/.github/workflows/test-nvidia-mlperf-inference-implementations.yml index 3ba2db04e..14a4c7265 100644 --- a/.github/workflows/test-nvidia-mlperf-inference-implementations.yml +++ b/.github/workflows/test-nvidia-mlperf-inference-implementations.yml @@ -24,7 +24,7 @@ jobs: exclude: - model: gptj-99.9 - system: phoenix1 - - system: GO-i9 + - system: GO-i91 steps: - name: Test MLPerf Inference NVIDIA ${{ matrix.model }} From 7ac5daf6f0bfea6bc1e4f0f44c8b52aa07ebff22 Mon Sep 17 00:00:00 2001 From: Arjun Suresh Date: Thu, 13 Feb 2025 09:09:50 +0000 Subject: [PATCH 2/8] Fix import in detect-sudo --- script/detect-sudo/customize.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/script/detect-sudo/customize.py b/script/detect-sudo/customize.py index dbfe5e5bb..96a16ad99 100644 --- a/script/detect-sudo/customize.py +++ b/script/detect-sudo/customize.py @@ -4,7 +4,7 @@ import select import sys import grp - +import threading def preprocess(i): From 599293b47c938e79eb555feebbd7a9f251235ec6 Mon Sep 17 00:00:00 2001 From: "github-actions[bot]" Date: Thu, 13 Feb 2025 09:10:13 +0000 Subject: [PATCH 3/8] [Automated Commit] Format Codebase [skip ci] --- script/detect-sudo/customize.py | 1 + 1 file changed, 1 insertion(+) diff --git a/script/detect-sudo/customize.py b/script/detect-sudo/customize.py index 96a16ad99..ff4de265f 100644 --- a/script/detect-sudo/customize.py +++ b/script/detect-sudo/customize.py @@ -6,6 +6,7 @@ import grp import threading + def preprocess(i): os_info = i['os_info'] From 8423065b9727d975f10fc99af4d3fe012ee216e0 Mon Sep 17 00:00:00 2001 From: Arjun Suresh Date: Thu, 13 Feb 2025 09:14:30 +0000 Subject: [PATCH 4/8] Fix import in detect-sudo --- script/detect-sudo/customize.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/script/detect-sudo/customize.py b/script/detect-sudo/customize.py index ff4de265f..c9d82816e 100644 --- a/script/detect-sudo/customize.py +++ b/script/detect-sudo/customize.py @@ -5,6 +5,7 @@ import sys import grp import threading +import getpass def preprocess(i): @@ -125,7 +126,6 @@ def prompt_sudo(): if os.geteuid() != 0 and not is_user_in_sudo_group(): # No sudo required for root user # Prompt for the password - import getpass if not os.isatty(sys.stdin.fileno()): print("Skipping password prompt - non-interactive terminal detected!") From b43f79c27eec620201cfbd7fd601f7d0f097d3f7 Mon Sep 17 00:00:00 2001 From: Arjun Suresh Date: Thu, 13 Feb 2025 12:17:46 +0000 Subject: [PATCH 5/8] Update test-nvidia-mlperf-inference-implementations.yml --- .../test-nvidia-mlperf-inference-implementations.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.github/workflows/test-nvidia-mlperf-inference-implementations.yml b/.github/workflows/test-nvidia-mlperf-inference-implementations.yml index 14a4c7265..2e144f0f5 100644 --- a/.github/workflows/test-nvidia-mlperf-inference-implementations.yml +++ b/.github/workflows/test-nvidia-mlperf-inference-implementations.yml @@ -58,7 +58,7 @@ jobs: pip install --upgrade mlcflow mlc pull repo mlcommons@mlperf-automations --branch=dev - mlcr --tags=run-mlperf,inference,_all-scenarios,_submission,_full,_r5.0-dev --preprocess_submission=yes --pull_changes=yes --pull_inference_changes=yes --execution_mode=valid --gpu_name=$gpu_name --pull_changes=yes --pull_inference_changes=yes --model=${{ matrix.model }} --submitter="MLCommons" --hw_name=$hw_name --implementation=nvidia --backend=tensorrt --category=$category --division=closed --docker_dt --docker_mlc_repo=mlcommons@mlperf-automations --docker_mlc_repo_branch=dev --adr.compiler.tags=gcc --device=cuda --use_model_from_host=yes --use_dataset_from_host=yes --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --clean $docker_string $submission_preprocessor_args --quiet + mlcr --tags=run-mlperf,inference,_all-scenarios,_submission,_full,_r5.0-dev --preprocess_submission=yes --pull_changes=yes --pull_inference_changes=yes --execution_mode=valid --gpu_name=$gpu_name --pull_changes=yes --pull_inference_changes=yes --model=${{ matrix.model }} --submitter="GATEOverflow " --hw_name=$hw_name --implementation=nvidia --backend=tensorrt --category=$category --division=closed --docker_dt --docker_mlc_repo=mlcommons@mlperf-automations --docker_mlc_repo_branch=dev --adr.compiler.tags=gcc --device=cuda --use_model_from_host=yes --use_dataset_from_host=yes --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --clean $docker_string $submission_preprocessor_args --quiet #mlcr --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/mlcommons/mlperf_inference_unofficial_submissions_v5.0 --repo_branch=auto-update --commit_message="Results from GH action on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name - mlcr --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/GATEOverflow/mlperf_inference_submissions_v5.0 --repo_branch=main --commit_message="Results from GH action on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name + mlcr --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/GATEOverflow/mlperf_inference_submissions_v5.0 --repo_branch=main --commit_message="Results from on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name From be575364c80e92cee3cafa6c7ca3bfff3fae9110 Mon Sep 17 00:00:00 2001 From: Arjun Suresh Date: Thu, 13 Feb 2025 13:36:54 +0000 Subject: [PATCH 6/8] Update test-nvidia-mlperf-inference-implementations.yml --- .../workflows/test-nvidia-mlperf-inference-implementations.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/test-nvidia-mlperf-inference-implementations.yml b/.github/workflows/test-nvidia-mlperf-inference-implementations.yml index 2e144f0f5..0cab69c89 100644 --- a/.github/workflows/test-nvidia-mlperf-inference-implementations.yml +++ b/.github/workflows/test-nvidia-mlperf-inference-implementations.yml @@ -60,5 +60,5 @@ jobs: mlcr --tags=run-mlperf,inference,_all-scenarios,_submission,_full,_r5.0-dev --preprocess_submission=yes --pull_changes=yes --pull_inference_changes=yes --execution_mode=valid --gpu_name=$gpu_name --pull_changes=yes --pull_inference_changes=yes --model=${{ matrix.model }} --submitter="GATEOverflow " --hw_name=$hw_name --implementation=nvidia --backend=tensorrt --category=$category --division=closed --docker_dt --docker_mlc_repo=mlcommons@mlperf-automations --docker_mlc_repo_branch=dev --adr.compiler.tags=gcc --device=cuda --use_model_from_host=yes --use_dataset_from_host=yes --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --clean $docker_string $submission_preprocessor_args --quiet #mlcr --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/mlcommons/mlperf_inference_unofficial_submissions_v5.0 --repo_branch=auto-update --commit_message="Results from GH action on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name - mlcr --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/GATEOverflow/mlperf_inference_submissions_v5.0 --repo_branch=main --commit_message="Results from on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name + mlcr --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/GATEOverflow/mlperf_inference_submissions_v5.0 --repo_branch=main --commit_message="Results from GH actions on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name From 75a4d11a86a95c18ff91c2a153a4eabd3c5f6e02 Mon Sep 17 00:00:00 2001 From: Arjun Suresh Date: Fri, 14 Feb 2025 04:47:22 +0530 Subject: [PATCH 7/8] Support mlperf checkpoint for llama2 --- script/get-ml-model-llama2/customize.py | 6 ++++++ script/get-ml-model-llama2/meta.yaml | 11 +++++++++++ 2 files changed, 17 insertions(+) diff --git a/script/get-ml-model-llama2/customize.py b/script/get-ml-model-llama2/customize.py index 7e8e0b4f1..494b7c9cf 100644 --- a/script/get-ml-model-llama2/customize.py +++ b/script/get-ml-model-llama2/customize.py @@ -29,6 +29,12 @@ def preprocess(i): else: if path == '' or not os.path.exists(path): env['MLC_TMP_REQUIRE_DOWNLOAD'] = 'yes' + if env['MLC_DOWNLOAD_SRC'] == "mlcommons": + i['run_script_input']['script_name'] = 'run-rclone' + if env.get('MLC_OUTDIRNAME', '') != '': + env['LLAMA2_CHECKPOINT_PATH'] = env['MLC_OUTDIRNAME'] + else: + env['LLAMA2_CHECKPOINT_PATH'] = os.getcwd() return {'return': 0} diff --git a/script/get-ml-model-llama2/meta.yaml b/script/get-ml-model-llama2/meta.yaml index 265b66925..0b0819678 100644 --- a/script/get-ml-model-llama2/meta.yaml +++ b/script/get-ml-model-llama2/meta.yaml @@ -19,6 +19,8 @@ prehook_deps: - enable_if_env: MLC_TMP_REQUIRE_DOWNLOAD: - 'yes' + MLC_DOWNLOAD_SRC: + - 'huggingface' env: {} extra_cache_tags: llama2,llama-2 force_env_keys: @@ -82,6 +84,15 @@ variations: MLC_ML_MODEL_PRECISION: int8 MLC_ML_MODEL_WEIGHT_DATA_TYPES: int8 group: precision + mlc: + group: download-source + default: true + env: + MLC_DOWNLOAD_SRC: mlcommons + hf: + group: download-source + env: + MLC_DOWNLOAD_SRC: huggingface meta-llama/Llama-2-70b-chat-hf: adr: hf-zoo: From ea708bb99041f89ce7f81a469546030f8fd00cae Mon Sep 17 00:00:00 2001 From: Arjun Suresh Date: Fri, 14 Feb 2025 04:47:49 +0530 Subject: [PATCH 8/8] Support mlperf checkpoint for llama2 --- script/get-ml-model-llama2/run-rclone.sh | 3 +++ 1 file changed, 3 insertions(+) create mode 100644 script/get-ml-model-llama2/run-rclone.sh diff --git a/script/get-ml-model-llama2/run-rclone.sh b/script/get-ml-model-llama2/run-rclone.sh new file mode 100644 index 000000000..2d7d39a15 --- /dev/null +++ b/script/get-ml-model-llama2/run-rclone.sh @@ -0,0 +1,3 @@ +rclone config create mlc-llama2 drive config_is_local=false scope=drive.readonly root_folder_id=11tBZvvrh0FCm3XuR5E849K42TqftYdUF +rclone config reconnect mlc-llama2: +rclone copy mlc-llama2:Llama-2-7b-chat-hf ${LLAMA2_CHECKPOINT_PATH}/Llama-2-7b-chat-hf -P