4.1 KiB

Raw Blame History

Train GPT with Colossal-AI

This example shows how to use Colossal-AI to run huggingface GPT training in distributed manners.

GPT

We use the GPT-2 model from huggingface transformers. The key learning goal of GPT-2 is to use unsupervised pre-training models to do supervised tasks.GPT-2 has an amazing performance in text generation, and the generated text exceeds people's expectations in terms of contextual coherence and emotional expression.

Requirements

Before you can launch training, you need to install the following requirements.

Install PyTorch

#conda
conda install pytorch==1.12.0 torchvision==0.13.0 torchaudio==0.12.0 cudatoolkit=11.3 -c pytorch
#pip
pip install torch==1.12.0+cu113 torchvision==0.13.0+cu113 torchaudio==0.12.0 --extra-index-url https://download.pytorch.org/whl/cu113

Install Colossal-AI v0.1.12 From Official Website

pip install colossalai==0.1.12+torch1.12cu11.3 -f https://release.colossalai.org

Install transformers

pip install transformers

This is just an example that we download PyTorch=1.12.0, CUDA=11.6 and colossalai=0.1.12+torch1.12cu11.3. You can download another version of PyTorch and its corresponding ColossalAI version. Just make sure that the version of ColossalAI is at least 0.1.10, PyTorch is at least 1.8.1 and transformers is at least 4.231. If you want to test ZeRO1 and ZeRO2 in Colossal-AI, you need to ensure Colossal-AI>=0.1.12.

Dataset

For simplicity, the input data is randonly generated here.

Training

bash run.sh

Pipeline Parallel

bash run_pp.sh

Training config

The train_gpt_demo.py provides three distributed plans, you can choose the plan you want in run.sh. The Colossal-AI leverages Tensor Parallel and Gemini + ZeRO DDP.

Colossal-AI
ZeRO1 (Colossal-AI)
ZeRO2 (Colossal-AI)
Pytorch DDP
Pytorch ZeRO

Performance

Testbed: a cluster of 8xA100 (80GB) and 1xAMD EPYC 7543 32-Core Processor (512 GB). GPUs are connected via PCI-e. ColossalAI version 0.1.13.

How dose Batch Size affect the efficency.

model	#GPU	policy	TP	batch per DP	Tflops
gpt2_10b	2	cpu	1	32	122.046
gpt2_10b	2	cpu	1	16	82.649
gpt2_10b	2	cpu	1	8	61.354

How dose the Placement Policy affect the efficency.

model	#GPU	policy	TP	batch per DP	Tflops
gpt2_10b	4	auto	1	8	88.657
gpt2_10b	4	cuda	1	8	OOM
gpt2_10b	4	cpu	1	8	61.354
gpt2_10b	4	const	1	8	82.137

How dose the Tensor Parallel Degree affect the efficency.

model	#GPU	policy	TP	batch per DP	Tflops
gpt2_10b	4	auto	1	8	88.657
gpt2_10b	4	auto	2	8	56.687
gpt2_10b	4	auto	4	8	29.019
gpt2_10b	4	auto	4	64	50.411
gpt2_20b	1	cpu	1	8	43.102
gpt2_20b	4	cpu	4	8	28.491

Touch the bar of model scale and batch size.

cpu is the most stable policy for large model and large batch size. One 8 GPU with TP=2, largest batch size of auto, const cpu is 64, 32 and 16, respectively.
Tensor parallel is necessary for 20B model to reduce model data memory requirement on each GPU.

model	#GPU	policy	TP	batch per DP	Tflops
gpt2_20b	4	cpu	1	64	CUDA OOM
gpt2_20b	4	auto	1/2	64	CUDA OOM
gpt2_20b	4	cpu	2	8	43.102
gpt2_20b	4	cpu	2	64	121.394
gpt2_20b	8	auto	2	16	99.871
gpt2_20b	8	cpu	2	64	125.170
gpt2_20b	8	const	2	32	105.415

model	#GPU	policy	TP	batch per DP	Tflops
gpt2_20b	8	cpu	2	8	46.895

4.1 KiB Raw Blame History