Skip to content

Commit

Permalink
Update llama2-model.libsonnet
Browse files Browse the repository at this point in the history
  • Loading branch information
@ManfeiBai
ManfeiBai authored Aug 21, 2023
1 parent c2a6695 commit 64d2765
Showing 1 changed file with 34 additions and 208 deletions.
242 changes: 34 additions & 208 deletions tests/pytorch/nightly/llama2-model.libsonnet
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,22 +19,6 @@ local tpus = import 'templates/tpus.libsonnet';
local utils = import 'templates/utils.libsonnet';

{
local llama2_google_next_inference_pretrained_models_eager = self.llama2_google_next_inference_pretrained_models_eager,
llama2_google_next_inference_pretrained_models_eager:: common.PyTorchTest {
local config = self,
modelName: 'l2-e',
paramsOverride:: {
scriptPath: 'llama/example_text_completion.py',
trainCommand: [
'torchrun --nproc_per_node 1',
self.scriptPath,
'--ckpt_dir llama_2_model/llama-2-13b-dummy/',
'--tokenizer_path llama_2_model/tokenizer.model',
'--max_seq_len 128 --max_batch_size 4',
],
},
command: self.paramsOverride.trainCommand,
},
local llama2_google_next_inference_pretrained_models = self.llama2_google_next_inference_pretrained_models,
llama2_google_next_inference_pretrained_models:: common.PyTorchTest {
local config = self,
Expand All @@ -44,207 +28,53 @@ local utils = import 'templates/utils.libsonnet';
trainCommand: [
'python3',
self.scriptPath,
'--ckpt_dir 7B/',
'--ckpt_dir llama/7B/',
'--tokenizer_path spiece.model',
'--max_seq_len 128 --max_batch_size 4',
],
},
command: self.paramsOverride.trainCommand,
},
local llama2_google_next_inference_fine_tuned_chat_models_eager = self.llama2_google_next_inference_fine_tuned_chat_models_eager,
llama2_google_next_inference_fine_tuned_chat_models_eager:: common.PyTorchTest {
local config = self,
modelName: 'l2-c-e',
paramsOverride:: {
scriptPath: 'llama/example_chat_completion.py',
trainCommand: [
'torchrun --nproc_per_node 1',
self.scriptPath,
'--ckpt_dir llama_2_model/llama-2-13b-dummy/',
'--tokenizer_path llama_2_model/tokenizer.model',
'--max_seq_len 512 --max_batch_size 4',
],
},
command: self.paramsOverride.trainCommand,
},
local llama2_google_next_inference_fine_tuned_chat_models = self.llama2_google_next_inference_fine_tuned_chat_models,
llama2_google_next_inference_fine_tuned_chat_models:: common.PyTorchTest {
local config = self,
modelName: 'l2-c',
paramsOverride:: {
scriptPath: 'llama/example_chat_completion.py',
trainCommand: [
'python3',
self.scriptPath,
'--ckpt_dir llama_2_model/llama-2-13b-dummy/',
'--tokenizer_path llama_2_model/tokenizer.model',
'--max_seq_len 512 --max_batch_size 4',
'--mp True --dynamo True',
],
},
command: self.paramsOverride.trainCommand,
},
local llama2_stable_tokenizer = self.llama2_stable_tokenizer,
llama2_stable_tokenizer:: common.PyTorchTest {
local config = self,
modelName: 'l2-t',
paramsOverride:: {
scriptPath: 'example.py',
trainCommand: [
'python3',
self.scriptPath,
'--ckpt_dir llama_2_model/llama-2-13b-dummy/',
'--tokenizer_path llama_2_model/tokenizer.model',
'--mp True --dynamo True',
],
},
command: self.paramsOverride.trainCommand,
},
local llama2_stable_quant = self.llama2_stable_quant,
llama2_stable_quant:: common.PyTorchTest {
local config = self,
modelName: 'l2-q',
paramsOverride:: {
scriptPath: 'example_xla.py',
trainCommand: [
'python3',
self.scriptPath,
'--tokenizer_path llama_2_model/tokenizer.model',
'--ckpt_dir llama_2_model/tokenizer.model',
'--max_seq_len 256',
'--max_batch_size 1',
'--temperature 0.8',
'--mp True --dynamo True',
],
},
command: self.paramsOverride.trainCommand,
},
local llama2_stable_quant_without_download = self.llama2_stable_quant_without_download,
llama2_stable_quant_without_download:: common.PyTorchTest {
local config = self,
modelName: 'l2-q-w',
paramsOverride:: {
scriptPath: 'example_xla.py',
trainCommand: [
'python3',
self.scriptPath,
'--tokenizer_path llama_2_model/tokenizer.model',
'--max_seq_len 256',
'--max_batch_size 1',
'--temperature 0.8',
'--dim 4096',
'--n_heads 32',
'--n_layers 32',
'--mp True',
],
},
command: self.paramsOverride.trainCommand,
},
local llama2_google_next_inference = self.llama2_google_next_inference,
llama2_google_next_inference:: common.PyTorchTpuVmMixin {
local pjrt = self.pjrt,
pjrt:: common.PyTorchTpuVmMixin {
modelName+: '-n-i',
tpuSettings+: {
tpuVmExports+: |||
export LD_LIBRARY_PATH=/usr/local/lib/
export PT_XLA_DEBUG=0
export USE_TORCH=ON
|||,
tpuVmExtraSetup: |||
sudo apt update
sudo apt-get -y install libopenblas-dev
pip install accelerate -U
sudo apt update
sudo apt-get -y install libopenblas-dev
pip3 uninstall -y torch torch_xla
## pip3 install https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch-nightly-cp38-cp38-linux_x86_64.whl
## pip3 install https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-nightly-cp38-cp38-linux_x86_64.whl
pip3 install https://storage.googleapis.com/tpu-pytorch/wheels/tpuvm/torch-nightly+20230422-cp38-cp38-linux_x86_64.whl
pip3 install https://storage.googleapis.com/tpu-pytorch/wheels/cuda/118/torch_xla-nightly+20230422-cp38-cp38-linux_x86_64.whl
pip3 install https://storage.googleapis.com/tpu-pytorch/wheels/cuda/118/torchvision-nightly+20230422-cp38-cp38-linux_x86_64.whl
pip3 uninstall -y libtpu-nightly
pip3 install torch_xla[tpuvm] --user
# show current path
pwd
## /home/xl-ml-test
ls
## pytorch snap
git clone -b llama2-google-next-inference https://github.com/pytorch-tpu/llama.git
pip3 uninstall torch torch_xla torchvision libtpu-nightly -y
sudo apt-get update -y
sudo apt-get install libomp5 -y
pip3 install mkl mkl-include
pip3 install tf-nightly tb-nightly tbp-nightly
pip3 install numpy
sudo apt-get install numactl -y
sudo apt-get install libopenblas-dev -y
pip3 install https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch-nightly+20230726-cp38-cp38-linux_x86_64.whl
pip3 install https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-nightly+20230726-cp38-cp38-linux_x86_64.whl
pip3 install torch_xla[tpuvm]
# install tokenizer model
wget https://storage.googleapis.com/tpu-pytorch/lsiyuan-experiment/llama/spiece.model
# git clone and build llama
git clone --branch llama2-google-next-inference https://github.com/pytorch-tpu/llama.git
cd llama
# show current path
pwd
## /home/xl-ml-test/llama
ls
## CODE_OF_CONDUCT.md CONTRIBUTING.md LICENSE MODEL_CARD.md README.md Responsible-Use-Guide.pdf USE_POLICY.md download.sh example_chat_completion.py
## example_text_completion.py llama requirements.txt reshard_checkpoints.py setup.py
pip list | grep torch
## torch 2.1.0
## torch-xla 2.1.0
## torchvision 0.16.0.dev20230817+cpu
pip install -r requirements.txt
pip install -e .
pip list | grep torch
## torch 2.1.0
## torch-xla 2.1.0
## torchvision 0.16.0.dev20230817+cpu
# prepare data
# show current path
pwd
## /home/xl-ml-test/llama
ls
## CODE_OF_CONDUCT.md CONTRIBUTING.md LICENSE MODEL_CARD.md README.md Responsible-Use-Guide.pdf USE_POLICY.md download.sh example_chat_completion.py
## example_text_completion.py llama llama.egg-info requirements.txt reshard_checkpoints.py setup.py
# --- wget https://storage.mtls.cloud.google.com/manfei_bucket/LLaMA2/llama_2_model.zip
# show current path
pwd
## /home/xl-ml-test/llama
ls
## CODE_OF_CONDUCT.md CONTRIBUTING.md LICENSE MODEL_CARD.md README.md Responsible-Use-Guide.pdf USE_POLICY.md download.sh example_chat_completion.py
## example_text_completion.py llama llama.egg-info llama_2_model.zip requirements.txt reshard_checkpoints.py setup.py
# --- sudo apt-get install unzip
# --- unzip llama_2_model.zip
## wget -nv -O llama_2_model.zip https://storage.mtls.cloud.google.com/manfei_bucket/LLaMA2/llama_2_model.zip
## unzip -o llama_2_model.zip
## 7B
gsutil cp gs://tpu-pytorch/lsiyuan-experiment/llama/spiece.model .
pip3 install -r requirements.txt
pip3 install -e .
# 7B config
mkdir 7B
cd 7B/
echo -e '{"dim": 4096, "multiple_of": 256, "n_heads": 32, "n_layers": 32, "norm_eps": 1e-05, "vocab_size": -1}' >> params.json
## unzip: cannot find zipfile directory in one of llama_2_model.zip or llama_2_model.zip.zip, and cannot find llama_2_model.zip.ZIP, period.
# show current path
cat params.json
pwd
## /home/xl-ml-test/llama
ls
## CODE_OF_CONDUCT.md CONTRIBUTING.md LICENSE MODEL_CARD.md README.md Responsible-Use-Guide.pdf USE_POLICY.md download.sh example_chat_completion.py
## example_text_completion.py llama llama.egg-info llama_2_model.zip requirements.txt reshard_checkpoints.py setup.py
|||,
},
},
local stable = self.stable,
stable:: common.PyTorchTpuVmMixin {
modelName+: '-s',
tpuSettings+: {
tpuVmExtraSetup: |||
git clone -b stable https://github.com/pytorch-tpu/llama.git
cd llama
pip install -r requirements.txt
pip install -e .
# prepare data
wget https://storage.mtls.cloud.google.com/manfei_bucket/LLaMA2/llama_2_model.zip
sudo apt-get install unzip
unzip llama_2_model.zip
|||,
},
},
local xla = self.xla,
xla:: common.PyTorchTpuVmMixin {
modelName+: '-xla',
tpuSettings+: {
tpuVmExtraSetup: |||
sudo apt update
sudo apt-get -y install libopenblas-dev
pip install accelerate -U
sudo apt update
sudo apt-get -y install libopenblas-dev
pip3 uninstall -y torch torch_xla
pip3 install https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch-nightly-cp310-cp310-linux_x86_64.whl
pip3 install https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-nightly-cp310-cp310-linux_x86_64.whl
pip3 uninstall -y libtpu-nightly
pip3 install torch_xla[tpuvm] --user
# set up env
echo 'export PATH=~/.local/bin:$PATH' >> ~/.bash_profile
|||,
},
},
Expand All @@ -255,10 +85,6 @@ local utils = import 'templates/utils.libsonnet';
},

configs: [
llama2_google_next_inference_pretrained_models + v4_8 + common.Functional + timeouts.Hours(3) + llama2_google_next_inference,
// llama2_google_next_inference_fine_tuned_chat_models + v4_8 + common.Functional + timeouts.Hours(3) + llama2_google_next_inference + xla,
llama2_stable_tokenizer + v4_8 + common.Functional + timeouts.Hours(3) + stable + xla,
llama2_stable_quant + v4_8 + common.Functional + timeouts.Hours(3) + stable + xla,
llama2_stable_quant_without_download + v4_8 + common.Functional + timeouts.Hours(3) + stable + xla,
llama2_google_next_inference_pretrained_models + v4_8 + common.Functional + timeouts.Hours(3) + pjrt,
],
}

0 comments on commit 64d2765

Please sign in to comment.