DEMix

This repository contains modeling utilities for "DEMix Layers: Disentangling Domains for Modular Language Modeling" (Gururangan et. al, 2021).

This code is a fork of Fairseq. It is based on Python 3.8, CUDA 11 and includes PyTorch 1.8.0, NCCL 2.8.4 and apex.

Dataset

The multidomain dataset scripts are housed in another repository, located here. Clone that repository and follow instructions to setup data to train on.

Follow that tutorial to generate data-bins on eight (small) example domains.

Make sure to set the DATA_DIR accordingly.

Fairseq Installation

If you've already made an environment from the dataset creation phase, just use that. Otherwise:

conda create env --name demix
cd demix/
pip install --editable .

Additionally, please make sure you have the dependencies above installed (check Fairseq documentation for more information).

Tutorial

Here we will follow a tutorial to train on the example domains from the tutorial in the DEMix-data repository. Note that the model that results from this tutorial is pretty bad, because we're working with very small amounts of data and also a small LM. This tutorial is there to help you quickly understand the pipeline, and ensure that each script completes successfully.

To replicate the DEMix paper, with a GPT-3 model, follow the instructions here.

Basic Training

After setting up the example domains, run the following to train a small language model. Note that the scripts in this paper assume you are running on a multi-node GPU cluster with SLURM.

First, allocate some nodes, with GPUs with at least 32GB of RAM. Here we allocate 1 node with 8 volta32GB GPUs.

salloc --gpus-per-node 8 --nodes 1  -C 'volta32gb' --ntasks-per-node 8 --cpus-per-task 10 --mem 400G --time XXX --partition YYY

Then run:

export NUM_GPUS=8
export DISTRIBUTED_PORT=12345
export MODEL=transformer_lm
export EXPERIMENT=demix
# $DATA_DIR was set in DEMix-data tutorial.
export DATA_BIN=${DATA_DIR}/data-bin/
export EXPERIMENT_SUFFIX=tutorial
export SERIALIZATION_DIR=$(pwd)/demix_tutorial_model
bash tutorial/train.sh $NUM_GPUS \
                    $DISTRIBUTED_PORT \
                    $MODEL \
                    $EXPERIMENT \
                    $DATA_BIN \
                    $SERIALIZATION_DIR \
                    $EXPERIMENT_SUFFIX

This will output a trained language model in ${SERIALIZATION_DIR}

To train balanced dense LM, set export EXPERIMENT=dense, to train unbalanced dense LM, set export EXPERIMENT=unbalanced, to train "+Domain Token" LM , set export EXPERIMENT=domain_token.

We have provided a simple script demix/train.sh, with the same interface, with all hyperparameter preset to help replicate results in the paper.

Evaluation

We have two ways to evaluate the demix language model: with and without mixing experts.

Evaluating without mixing experts

To evaluate the language model without mixing experts, you can supply the checkpoint from a GPU on a particular rank (to specify the use of the domain expert that was trained on that GPU):

export DATA_BIN=${DATA_DIR}/data-bin/
export GPU_RANK=0
export PATH_TO_CHECKPOINT=${SERIALIZATION_DIR}/checkpoint_last-rank-${GPU_RANK}.pt
export OUTPUT_PATH=eval_output.jsonl
export SPLIT=valid
export DOMAIN=imdb
bash tutorial/eval_lm.sh $DATA_BIN $PATH_TO_CHECKPOINT $OUTPUT_PATH $SPLIT $DOMAIN

To evaluate on test data, set export SPLIT=test

The same script is used for the other baselines.

For the +domain token model, you can additionally supply a domain token to use at test time:

export DOMAIN_TOKEN=XXX
bash tutorial/eval_lm.sh $DATA_BIN $PATH_TO_CHECKPOINT $OUTPUT_PATH $SPLIT $DOMAIN $DOMAIN_TOKEN

Evaluating with mixing experts

First, we estimate the posterior distribution on 100 sequences of validation data of the domain using the following command:

export DATA_BIN=${DATA_DIR}/data-bin
export DOMAIN=imdb
export DEV_POSTERIOR_OUTPUT=dev_posteriors.jsonl
# set NUM_EVALUATION_GPUS equal to the number of experts you'd like to ensemble.
export NUM_EVALUATION_GPUS=8;
bash tutorial/mix_eval_lm.sh $NUM_EVALUATION_GPUS $DATA_BIN  ${SERIALIZATION_DIR}/checkpoint_last-rank-0.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-1.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-2.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-3.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-4.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-5.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-6.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-7.pt $DOMAIN $DEV_POSTERIOR_OUTPUT estimate;

Then, we open $POSTERIOR_OUTPUT, extracting the exp_avg_posterior value of the last line in that file:

export POSTERIOR=$(tail -n 1 $DEV_POSTERIOR_OUTPUT | jq -rc '.exp_avg_posterior | join(",")')

We use this posterior as the domain prior (supplied as a string) when evaluating on test data, like so:

bash tutorial/mix_eval_lm.sh $NUM_EVALUATION_GPUS $DATA_BIN  ${SERIALIZATION_DIR}/checkpoint_last-rank-0.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-1.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-2.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-3.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-4.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-5.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-6.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-7.pt $DOMAIN $DEV_POSTERIOR_OUTPUT eval $POSTERIOR cached_prior;

Adapting the Language Model

We additionally provide scripts to adapt the language model to a new domain.

DEMix DAPT

In this tutorial, we just adapt one of the existing experts to a new example domain in the demix-data project, located in /path/to/demix-data/new_example_domains.

First, we need to figure out which domain expert has the most affinity to the target domain we want to adapt to:

export NEW_DATA_BIN=/private/home/suching/demix-data/new_example_domains/data-bin/
export NEW_DOMAIN=acl_papers
export DEV_POSTERIOR_OUTPUT=${NEW_DOMAIN}_posterior.jsonl
# set NUM_EVALUATION_GPUS equal to the number of experts you'd like to ensemble.
export NUM_EVALUATION_GPUS=8;
bash tutorial/mix_eval_lm.sh $NUM_EVALUATION_GPUS $NEW_DATA_BIN  ${SERIALIZATION_DIR}/checkpoint_last-rank-0.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-1.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-2.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-3.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-4.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-5.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-6.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-7.pt $NEW_DOMAIN $DEV_POSTERIOR_OUTPUT estimate;
export POSTERIOR=$(tail -n 1 $DEV_POSTERIOR_OUTPUT | jq -rc '.exp_avg_posterior | join(",")')

Here, we find that the most likely expert is expert number 5.

export POSTERIOR=$(tail -n 1 $DEV_POSTERIOR_OUTPUT | jq -rc '.exp_avg_posterior | join(",")')
echo $POSTERIOR

We then adapt expert 5 to the target domain using the tutorial/dapt.sh script, using DEMix DAPT:

export PATH_TO_CHECKPOINT=${SERIALIZATION_DIR}/checkpoint_last-rank-5.pt
export UNFREEZE_PARAMETERS=feedforward
export NEW_SERIALIZATION_DIR=$(pwd)/${NEW_DOMAIN}_demix_dapt
export EXPERIMENT_SUFFIX=test
bash tutorial/dapt.sh $NEW_DATA_BIN $NEW_DOMAIN $PATH_TO_CHECKPOINT $UNFREEZE_PARAMETERS $NEW_SERIALIZATION_DIR $EXPERIMENT_SUFFIX

Once this is trained, you can add that expert to your ensemble when evaluating on new data:

export NEW_DATA_BIN=/path/to/demix-data/new_example_domains/data-bin/
export NEW_DOMAIN=acl_papers
export DEV_POSTERIOR_OUTPUT=${NEW_DOMAIN}_posterior.jsonl
# set NUM_EVALUATION_GPUS equal to the number of experts you'd like to ensemble.
export NUM_EVALUATION_GPUS=8;
export PATH_TO_NEW_EXPERT=${NEW_SERIALIZATION_DIR}/checkpoint_last-rank-0.pt
bash tutorial/mix_eval_lm.sh $NUM_EVALUATION_GPUS $NEW_DATA_BIN  ${SERIALIZATION_DIR}/checkpoint_last-rank-0.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-1.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-2.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-3.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-4.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-5.pt:${SERIALIZATION_DIR}/checkpoint_last-rank-6.pt:${PATH_TO_NEW_EXPERT} $NEW_DOMAIN $DEV_POSTERIOR_OUTPUT estimate;
export POSTERIOR=$(tail -n 1 $DEV_POSTERIOR_OUTPUT | jq -rc '.exp_avg_posterior | join(",")')

Dense DAPT

If you wanted to do Dense DAPT instead, just change the environment variables:

export PATH_TO_CHECKPOINT=/path/to/dense/model/checkpoint_last.pt
export FEEDFORWARD_OR_FULL=full
export SERIALIZATION_DIR=$(pwd)/${NEW_DOMAIN}_dense_dapt
export EXPERIMENT_SUFFIX=test
bash tutorial/dapt.sh $NEW_DATA_BIN $NEW_DOMAIN $PATH_TO_CHECKPOINT $FEEDFORWARD_OR_FULL $SERIALIZATION_DIR $EXPERIMENT_SUFFIX

DEMix Layers for Modular Language Modeling

Related tags

Overview

DEMix

Dataset

Fairseq Installation

Tutorial

Basic Training

Evaluation

Evaluating without mixing experts

Evaluating with mixing experts

Adapting the Language Model

DEMix DAPT

Dense DAPT

Owner

Suchin

Omniverse sample scripts - A guide for developing with Python scripts on NVIDIA Ominverse

Stream images from a connected camera over MQTT, view using Streamlit, record to file and sqlite

Official code for our CVPR '22 paper "Dataset Distillation by Matching Training Trajectories"

FastReID is a research platform that implements state-of-the-art re-identification algorithms.

Brain tumor detection using CNN (InceptionResNetV2 Model)

Awesome Weak-Shot Learning

The code uses SegFormer for Semantic Segmentation on Drone Dataset.

Official implementation of the ICCV 2021 paper "Conditional DETR for Fast Training Convergence".

MakeItTalk: Speaker-Aware Talking-Head Animation

PyTorch implementation of SMODICE: Versatile Offline Imitation Learning via State Occupancy Matching

A Streamlit component to render ECharts.

Implementation of self-attention mechanisms for general purpose. Focused on computer vision modules. Ongoing repository.

A repo that contains all the mesh keys needed for mesh backend, along with a code example of how to use them in python

Geometric Algebra package for JAX

Code release for NeRF (Neural Radiance Fields)

PyTorch - Python + Nim

PerfFuzz: Automatically Generate Pathological Inputs for C/C++ programs

Capture all information throughout your model's development in a reproducible way and tie results directly to the model code!

The code for the NSDI'21 paper "BMC: Accelerating Memcached using Safe In-kernel Caching and Pre-stack Processing".

DeOldify - A Deep Learning based project for colorizing and restoring old images (and video!)