Deduplicating Training Data Makes Language Models Better

Overview

Deduplicating Training Data Makes Language Models Better

This repository contains code to deduplicate language model datasets as descrbed in the paper "Deduplicating Training Data Makes Language Models Better" by Katherine Lee, Daphne Ippolito, Andrew Nystrom, Chiyuan Zhang, Douglas Eck, Chris Callison-Burch and Nicholas Carlini. This repository contains both the ExactSubstr deduplication implementation (written in Rust) along with the scripts we used in the paper to perform deduplication and inspect the results (written in Python). In an upcoming update, we will add files to reproduce the NearDup-deduplicated versions of the C4, RealNews, LM1B, and Wiki-40B-en datasets.

This is not an officially supported Google product.

Why deduplicate?

When datasets are created by scraping raw text from the Internet, this will often result in the same sequences being repeated multiple times (e.g., we find a single 50 word sequence that is repeated in the C4 dataset 60,000 times). Training models on deduplicated datasets is faster (because they see fewer total examples) and experimentally results in models with similar or better perplexity to models trained on data that hasn't been deduplicated. Moreover, language models are less likely to exhibit memorization when their training data has been well-deduplicated.

Citing this work

If you use this repository or our deduplicated datasets you can cite

@article{lee2021deduplicating,
      title={Deduplicating Training Data Makes Language Models Better}, 
      author={Katherine Lee and Daphne Ippolito and Andrew Nystrom and Chiyuan Zhang and Douglas Eck and Chris Callison-Burch and Nicholas Carlini},
      journal={arXiv preprint arXiv:2107.06499},
      year={2021},
}

Exact Deduplication Code

We provide an implementation of the exact deduplication technique used in the paper. This is very much research code. It is (a very slightly cleaned up) version of exactly what we do in the paper. It assumes that you want to deduplicate something the size of C4 (~300GB) running on a machine with 96 cores and >600GB of RAM. If you only want to use this for reasonably-sized datasets, you should change the number of parallel threads from 96 to something smaller. If your machine is big enough, there should be no upper bound on the size of the dataset it can handle (well, 2^64-1 bytes is the limit, but I think we can all agree that's essentially unlimited).

We build a suffix array (based on Andrew Gallant's suffix array implementation) in src/table.rs. It has some minor changes from the original version that make it so we can't just import this library as a crate. First, we need 64-bit integers. The original implementation says that u32 works for "reasonably sized documents (~4GB)" but we're working with unreasonably sized documents. So we need u64. Second, we don't want UTF8 strings. Everything is a [u8] byte array, because we might be working over token sequences which aren't valid UTF8. The main complication in the rest of src/main.rs is the fact that we want things to run in parallel, and we probably can't fit the entire suffix array into memory. And so all of our algorithms are designed around these constraints.

If you just want to run the rust deduplicator, then you will only need to install Rust:

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

If you additionally want to generate datasets to run the rust script on (and you probably do) then you will need python dependencies:

pip3 install numpy scipy tensorflow tensorflow_datasets transformers sentencepiece

Basic Usage

If you just want to reproduce the result of this paper, or deduplicate any language model that's already in the Tensorflow Datasets (TFDS) format, then you can just run the following commands:

cargo build

to compile the rust code, and then run

python3 scripts/load_dataset.py --data_dir $LOAD_DIR --save_dir $SAVE_DIR --name $DATASET --split $SPLIT [--tokenize]

For example, to get the LM1B training set you could run python3 scripts/load_dataset.py --data_dir ~/tensorflow_datasets --save_dir data --name lm1b --split test. This should will take just a few seconds to run on the test set or about an hour if running with the train set instead.

If the dataset is really big, you might want to add the --tokenize flag. This will shrink the dataset by roughly a factor of two by tokenizing it with the GPT-2 tokenizer.

And then to construct the suffix array run

python3 scripts/make_suffix_array.py [path/to/dataset]

For example, if you run python3 scripts/make_suffix_array.py data/lm1b.test, this will create a file data/lm1b.test.table.bin containing the suffix array. Again, this should be fast, about two hours on the LM1B train set when run single-thread and a few minutes on 96 cores.

(If you get an error that you have too many open files, that's because this script opens lots of files. You should run ulimit -Sn 1000000 to "fix" the error.)

Querying a suffix array to find duplicated examples

Start by loading and building a suffix array for a dataset as described above

Once you have the suffix array, you now query the dataset to find all occurances of a particular string. To do this, run

python3 scripts/count_occurances.py --suffix [path/to/suffix_array] [--query query_string] [--query_file /path/to/query]

On the LM1B test set, running python3 scripts/count_occurances.py --suffix data/lm1b.test --query " on Tuesday" should return 1288. If you tokenized the dataset, then you should pass --tokenizetocount_occurences.py` as well, to get the same result (plus or minus tokenization differences).

If you want to confirm this the outputted number is correct (assuming you haven't tokenized), you can run cat /tmp/lm1b.test | grep -ao " on Tuesday" and get the same result.

Advanced Usage

The above scripts work by calling into the core Rust suffix array deduplicator. If you want to do each step yourself, the following options are available:

Single threaded suffix array construction

To build a suffix array for any particular file, you can run

cargo run save [file_path]

This will create a file called [file_path].table.bin which contains the suffix array for the file provided. This algorithm is linear time, but (a) only runs on a single core, and (b) has memory requirement O(big * len(file)) which is prohibitive for large files.

Parallel suffix array construction

To build a suffix array for an extremely large file (e.g., ~about as much RAM as available) it is better to run the script

python scripts/make_suffix_array.py [file_path]

This script will build the suffix array in parallel by splitting the single file into chunks, generating suffix arrays for each chunk, and then merging the suffix arrays together to form the full suffix array. Note that in general this algorithm is quadratic, but when the maximum substring length is short relative to the total file length (as it is, when generating suffix arrays for N independent training examples) it will never reach this worst case behavior.

The two steps are described below.

Building a piece of a suffix array from a piece of a file

The first generats a suffix array from a piece of a file. This is implemented by running

cargo run save_part [file_path] [byte_start] [byte_end]

And builds a suffix array for the byte sequence between [byte_start] and [byte_end] for the given file. Multiple of these can be run in parallel to build a suffix array for a file quickly.

Merging suffix array pieces to create a single suffix array

Given the several independent suffix arrays, merging them is now just a matter of calling

cargo run merge_parallel [path_to_partial_suffix_trees,...] [tmp_output_directory]

to generate a collection of ordered suffix arrays pieces in the output directory. The final step just requires merging these together

cat [tmp_output_directory]/* > [file_path].table.bin

Finding Duplicates

Given a suffix array file, as generated in the prevous section, it can now be queried for interesting statistics. The simplest operation, counting occurrences of particular substrings, takes O(log(N)) time and O(query_length) memory requirements, (as shown above with scripts/count_occurances.py). To do this you can run:

cargo run count_occurances /path/to/dataset /path/to/query_file

(Indeed, the python script is just a wrapper that makes calling this nicer, with the option for tokenization.) This is useful mainly as a commandline interface to interact with the dataset to find interesting properties. To run more sophisticated analysis, use the tools described below:

Finding duplicates between two documents

Given a document A and another document B, we can find all duplicates betwen the two by (1) constructing suffix arrays for both, and then (2) linearly walking the suffix arrays in order to find all duplicates of a given length.

Once the suffix array for the dataset has been constructed, this algorithm therefore requires time O(len(dataset) + len(query)) and space O(len(dataset)). It is better to run this algorithm when the number of queries into the dataset is greater than O(len(dataset)/log(len(query))). However note that the prior code requires disk seeks and and this implementation is a linear scan through the suffix array table, so in practice there is at least a factor-of-10 speedup here. As a rough order of magnitude, for a dataset with ~100GB, it is faster to run similar_parallel when querying with more than a few megabytes of text. Otherwise it is probably faster to run count_occurances.

Notice that this command also requires that the entire dataset fits in memory. For many datasets this is not a problem, but the C4 dataset is 350 GB and the Pile dataset is 750 GB (both even after tokenization). The machine must therefore have a lot of RAM for this to work.

cargo run similar_parallel [dataset1] [dataset2]

This creates lots of containing the position of all examples in dataset2 that are also in dataset1. (The code could also do the inverse at the same time, if you want to modify it slightly.) However it spits this out in some not-very-useful form: a list of tokens x_i so that dataset2[x_i:x_i+100] is also in dataset1. But this probably has overlaps.

The second step is then to run

cargo run collect_similar [dataset2]. This converts the result to instead compute ranges so that instead we have dataset2[xi:yi] match.

Finding duplicates within one document

To find duplicates that are contained within one document (for example, to actually deduplicate a dataset as we do in the paper) run the command

cargo run selfsimilar_parallel [dataset]

This will find all repeated substrings contained in the dataset above a given length threshold. Again run collect_similar to find the indexs of repeated examples.

Approx Deduplication Results

Coming soon.

Owner
Google Research
Google Research
A standard framework for modelling Deep Learning Models for tabular data

PyTorch Tabular aims to make Deep Learning with Tabular data easy and accessible to real-world cases and research alike.

801 Jan 08, 2023
Improving Object Detection by Estimating Bounding Box Quality Accurately

Improving Object Detection by Estimating Bounding Box Quality Accurately Abstrac

2 Apr 14, 2022
Complex-Valued Neural Networks (CVNN)Complex-Valued Neural Networks (CVNN)

Complex-Valued Neural Networks (CVNN) Done by @NEGU93 - J. Agustin Barrachina Using this library, the only difference with a Tensorflow code is that y

youceF 1 Nov 12, 2021
This repository contains several jupyter notebooks to help users learn to use neon, our deep learning framework

neon_course This repository contains several jupyter notebooks to help users learn to use neon, our deep learning framework. For more information, see

Nervana 92 Jan 03, 2023
My implementation of Fully Convolutional Neural Networks in Keras

Keras-FCN This repository contains my implementation of Fully Convolutional Networks in Keras (Tensorflow backend). Currently, semantic segmentation c

The Duy Nguyen 15 Jan 13, 2020
TiP-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modeling

TiP-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modeling This is the official code release for the paper 'TiP-Adapter: Training-fre

peng gao 189 Jan 04, 2023
ProFuzzBench - A Benchmark for Stateful Protocol Fuzzing

ProFuzzBench - A Benchmark for Stateful Protocol Fuzzing ProFuzzBench is a benchmark for stateful fuzzing of network protocols. It includes a suite of

155 Jan 08, 2023
Code for Multimodal Neural SLAM for Interactive Instruction Following

Code for Multimodal Neural SLAM for Interactive Instruction Following Code structure The code is adapted from E.T. and most training as well as data p

7 Dec 07, 2022
Cold Brew: Distilling Graph Node Representations with Incomplete or Missing Neighborhoods

Cold Brew: Distilling Graph Node Representations with Incomplete or Missing Neighborhoods Introduction Graph Neural Networks (GNNs) have demonstrated

37 Dec 15, 2022
Deep learning for Engineers - Physics Informed Deep Learning

SciANN: Neural Networks for Scientific Computations SciANN is a Keras wrapper for scientific computations and physics-informed deep learning. New to S

SciANN 195 Jan 03, 2023
Visual odometry package based on hardware-accelerated NVIDIA Elbrus library with world class quality and performance.

Isaac ROS Visual Odometry This repository provides a ROS2 package that estimates stereo visual inertial odometry using the Isaac Elbrus GPU-accelerate

NVIDIA Isaac ROS 343 Jan 03, 2023
Learning to Draw: Emergent Communication through Sketching

Learning to Draw: Emergent Communication through Sketching This is the official code for the paper "Learning to Draw: Emergent Communication through S

19 Jul 22, 2022
HAR-stacked-residual-bidir-LSTMs - Deep stacked residual bidirectional LSTMs for HAR

HAR-stacked-residual-bidir-LSTM The project is based on this repository which is presented as a tutorial. It consists of Human Activity Recognition (H

Guillaume Chevalier 287 Dec 27, 2022
Parsing, analyzing, and comparing source code across many languages

Semantic semantic is a Haskell library and command line tool for parsing, analyzing, and comparing source code. In a hurry? Check out our documentatio

GitHub 8.6k Dec 28, 2022
Semi-supervised Implicit Scene Completion from Sparse LiDAR

Semi-supervised Implicit Scene Completion from Sparse LiDAR Paper Created by Pengfei Li, Yongliang Shi, Tianyu Liu, Hao Zhao, Guyue Zhou and YA-QIN ZH

114 Nov 30, 2022
A clean and robust Pytorch implementation of PPO on continuous action space.

PPO-Continuous-Pytorch I found the current implementation of PPO on continuous action space is whether somewhat complicated or not stable. And this is

XinJingHao 56 Dec 16, 2022
🌳 A Python-inspired implementation of the Optimum-Path Forest classifier.

OPFython: A Python-Inspired Optimum-Path Forest Classifier Welcome to OPFython. Note that this implementation relies purely on the standard LibOPF. Th

Gustavo Rosa 30 Jan 04, 2023
This repository provides code for "On Interaction Between Augmentations and Corruptions in Natural Corruption Robustness".

On Interaction Between Augmentations and Corruptions in Natural Corruption Robustness This repository provides the code for the paper On Interaction B

Meta Research 33 Dec 08, 2022
subpixel: A subpixel convnet for super resolution with Tensorflow

subpixel: A subpixel convolutional neural network implementation with Tensorflow Left: input images / Right: output images with 4x super-resolution af

Atrium LTS 2.1k Dec 23, 2022
ReConsider is a re-ranking model that re-ranks the top-K (passage, answer-span) predictions of an Open-Domain QA Model like DPR (Karpukhin et al., 2020).

ReConsider ReConsider is a re-ranking model that re-ranks the top-K (passage, answer-span) predictions of an Open-Domain QA Model like DPR (Karpukhin

Facebook Research 47 Jul 26, 2022