Skip to content

Latest commit

 

History

History
 
 

llama

README.md

Llama 70B for Intel® Gaudi® MLPerf™ Inference Submission

This directory provides instructions to reproduce Intel Gaudi's results for MLPerf™ Inference submission.
MLPerf™ is a trademark and service mark of MLCommons Association in the United States and other countries.
All rights reserved. Unauthorized use is strictly prohibited.

Setup

Please follow the instructions provided in the Intel® Gaudi® Installation Guide to set up the environment.

CPU Performance Mode

Set CPU to performance mode:

echo performance | sudo tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

In order to verify the CPU mode, run:

cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

Huge Pages

It is recommended to set the number of huge pages as provided below:

#set current hugepages
sudo sysctl -w vm.nr_hugepages=100000
#Remove old entry if exists in sysctl.conf
sudo sed --in-place '/nr_hugepages/d' /etc/sysctl.conf
#Insert huge pages settings to persist
echo "vm.nr_hugepages=100000" | sudo tee -a /etc/sysctl.conf

Clone Intel Gaudi Model-References

Clone this repository and switch to the branch that matches your Intel Gaudi software version. You can run the hl-smi utility to determine the Intel Gaudi software version.

git clone --recurse-submodules -b [Intel Gaudi software version] https://github.com/HabanaAI/Model-References

Prepare Intel-HabanaLabs MLPerf Inference Container

export INTEL_HABANALABS_DIR=$PWD/Model-References/MLPERF4.0/Inference
docker run --privileged --security-opt seccomp=unconfined   \
           --name mlperf-intel-habanalabs -td               \
           -v /dev:/dev                                     \
           --device=/dev:/dev                               \
           -v /sys/kernel/debug:/sys/kernel/debug           \
           -v /tmp:/tmp                                     \
           -v $INTEL_HABANALABS_DIR:/root/Intel-HabanaLabs/ \
           --cap-add=sys_nice --cap-add=SYS_PTRACE          \
           --user root --workdir=/root --net=host           \
           --ulimit memlock=-1:-1 vault.habana.ai/gaudi-docker-mlperf/ver4.0/pytorch-installer-2.1.1:1.14.98-33
docker exec -it mlperf-intel-habanalabs bash

Get the Model

Choose one of the two available paths for downloading the model.

MLCommons Members Download

MLCommons hosts the model and preprocessed dataset for download exclusively by MLCommons Members. You must first agree to the confidentiality notice, then follow the link to a directory containing Rclone download instructions.

External Download

Using an e-mail registered for HuggingFace (if you do not have an account, you will need to create one), go to llama2-request-link and make a request. Having it accepted, go to HuggingFace Llama2 page and ask for an access to the model. Please note your HuggingFace authentication credentials as you may be required to provide them when cloning the below. The download requires Git Large Files Storage:

mkdir -p /mnt/weka/data/pytorch/llama2/
pushd /mnt/weka/data/pytorch/llama2/
apt-get update
apt-get install git-lfs
git clone https://huggingface.co/meta-llama/Llama-2-70b-chat-hf Llama-2-70b-chat-hf
popd

Get the Dataset

pushd /root/Intel-HabanaLabs/llama
export EXPORT_DIR=${PWD}/open_orca
mkdir -p /mnt/weka/data/mlperf_inference/llama2/
export DATASET_PATH=/mnt/weka/data/mlperf_inference/llama2/processed-data.pkl

curl https://rclone.org/install.sh | bash
rclone config create mlc-inference s3 provider=Cloudflare access_key_id=f65ba5eef400db161ea49967de89f47b \
secret_access_key=fbea333914c292b854f14d3fe232bad6c5407bf0ab1bebf78833c2b359bdfd2b \
endpoint=https://c2686074cb2caf5cbaf6d134bdba8b47.r2.cloudflarestorage.com
rclone copy mlc-inference:mlcommons-inference-wg-public/open_orca ./open_orca -P

pushd open_orca
gzip -d open_orca_gpt4_tokenized_llama.sampled_24576.pkl.gz
popd

mv ${EXPORT_DIR}/open_orca_gpt4_tokenized_llama.sampled_24576.pkl ${DATASET_PATH}
md5sum ${DATASET_PATH}
popd

The md5sum of generated dataset file should be 5fe8be0a7ce5c3c9a028674fd24b00d5.

Reproduce Results

99 and 99.9 Accuracy

The same script was submitted for both 99 and 99.9 benchmarks - no additional improvements were made for low accuracy (99), and 99.9 results were used for 99 as well.

Get Started

Source the necessary files:

cd /root/Intel-HabanaLabs
source functions.sh

Generate Results

To generate full submission results, run the following command:

build_mlperf_inference --output-dir <path_to_output_dir> --submission llama-99.9-fp8

The command produces results from accuracy and performance runs for both Offline and Server scenarios. Logs can be found under /output_dir/logs/model/, e.g. /results/logs/llama-99.9-fp8/

To generate results for Offline and Server scenarios separately, run the following commands:

build_mlperf_inference --output-dir <path_to_output_dir> --submission llama-99.9-fp8_Offline
build_mlperf_inference --output-dir <path_to_output_dir> --submission llama-99.9-fp8_Server

Logs can be found under /output_dir/logs/model/scenario/, e.g. /results/logs/llama-99.9-fp8/Offline/

To generate results for accuracy and performance separately, add --mode flag as in one of the following commands:

build_mlperf_inference --output-dir <path_to_output_dir> --submission llama-99.9-fp8_Server --mode acc
build_mlperf_inference --output-dir <path_to_output_dir> --submission llama-99.9-fp8_Offline --mode perf

Logs can be found under /output_dir/logs/model/scenario/mode/, e.g. /results/logs/llama-99.9-fp8/Offline/accuracy/

Performance Optimization with FP8 Flow

All linear operators' input activations and weights (linear and matmul operators) are quantized to FP8-143. The weights are pre-quantized to FP8-143.

Calibration

The submission contains measurement files required for FP8 quantization.

The following procedure was used to generate them:

cd /root/Intel-HabanaLabs/llama
export QUANT_CONFIG=hqt/llama2-70b-8x/config_meas_maxabs.json
deepspeed --num_gpus 8 llama_greedy.py --model_name_or_path /mnt/weka/data/pytorch/llama2/Llama-2-70b-chat-hf/ \
  --bf16 --attn_softmax_bf16 --use_hpu_graphs --use_kv_cache --batch_size 128 --reuse_cache                    \
  --trim_logits --limit_hpu_graphs --dataset $EXPORT_DIR/open_orca_gpt4_tokenized_llama.calibration_1000.pkl

The Quantization Toolkit is described in the Intel Gaudi documentation.

Supported Configurations

Validated on Intel Gaudi Software Version Framework Version(s) Mode
Gaudi 2 1.17.1 PyTorch 2.3.1 Inference