-
Notifications
You must be signed in to change notification settings - Fork 68
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
High level fuzzy duplicates module (#46)
* Initial pass at fuzzy dedup api Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Update deprecated shuffle arg Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * dask_cuda gpu only import Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Move fuzzy_dedup imports to optional Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * more tests Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Move FuzzyDeDupConfig to it's own class Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Add example script and config file, fix typo Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Remove slurm examples for gpu dedup Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Add config module Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Rename FuzzyDeDupConfig and minhash_length to FuzzyDuplicatesConfig, num_hashes Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Add comments and update example Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> * Write to same format as input in fuzzy dedup example Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com> --------- Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>
- Loading branch information
Showing
19 changed files
with
670 additions
and
477 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,16 @@ | ||
cache_dir: "./fuzzy_dedup_cache" | ||
# Optional Params below with default values | ||
# profile_dir: null | ||
# id_field: "id" | ||
# text_field: "text" | ||
|
||
# seed: 42 | ||
# char_ngrams: 5 | ||
# num_buckets: 20 | ||
# hashes_per_bucket: 13 | ||
# use_64_bit_hash: false | ||
# buckets_per_shuffle: 1 | ||
|
||
# false_positive_check: True | ||
# num_anchors: 2 | ||
# jaccard_threshold: 0.8 |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,109 @@ | ||
# Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. | ||
# | ||
# Licensed under the Apache License, Version 2.0 (the "License"); | ||
# you may not use this file except in compliance with the License. | ||
# You may obtain a copy of the License at | ||
# | ||
# http://www.apache.org/licenses/LICENSE-2.0 | ||
# | ||
# Unless required by applicable law or agreed to in writing, software | ||
# distributed under the License is distributed on an "AS IS" BASIS, | ||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
# See the License for the specific language governing permissions and | ||
# limitations under the License. | ||
|
||
import argparse | ||
import time | ||
|
||
import dask | ||
from dask import dataframe as dd | ||
|
||
from nemo_curator import FuzzyDuplicates, FuzzyDuplicatesConfig | ||
from nemo_curator.datasets import DocumentDataset | ||
from nemo_curator.utils.distributed_utils import get_client, write_to_disk | ||
from nemo_curator.utils.script_utils import add_distributed_args | ||
|
||
|
||
def pre_imports(): | ||
import cudf # noqa: F401 | ||
|
||
|
||
def main(args): | ||
|
||
dataset_dir = "/path/to/dataset" | ||
log_dir = "./" | ||
cache_dir = "./fuzzy_cache" | ||
output_dir = "./output" | ||
dataset_id_field = "id" | ||
dataset_text_field = "text" | ||
|
||
filetype = "parquet" | ||
|
||
# Fuzzy dup calculation only supports the cuDF/GPU backend | ||
backend = "cudf" | ||
assert args.device == "gpu" | ||
|
||
with dask.config.set({"dataframe.backend": backend}): | ||
client = get_client(args, args.device) | ||
client.run(pre_imports) | ||
|
||
t0 = time.time() | ||
if filetype == "parquet": | ||
input_dataset = DocumentDataset( | ||
dd.read_parquet( | ||
dataset_dir, | ||
columns=[dataset_id_field, dataset_text_field], | ||
blocksize="256MiB", | ||
aggregate_files=True, | ||
) | ||
) | ||
elif filetype == "jsonl": | ||
input_dataset = DocumentDataset.read_json( | ||
dataset_dir, | ||
backend=backend, | ||
) | ||
|
||
fuzzy_dedup_config = FuzzyDuplicatesConfig( | ||
cache_dir=cache_dir, | ||
id_field=dataset_id_field, | ||
text_field=dataset_text_field, | ||
seed=42, | ||
char_ngrams=5, | ||
num_buckets=20, | ||
hashes_per_bucket=13, | ||
use_64_bit_hash=False, | ||
buckets_per_shuffle=5, | ||
false_positive_check=True, | ||
num_anchors=2, | ||
jaccard_threshold=0.8, | ||
) | ||
fuzzy_dup = FuzzyDuplicates(logger=log_dir, config=fuzzy_dedup_config) | ||
duplicates = fuzzy_dup(dataset=input_dataset) | ||
|
||
# By default all duplicate id's and the group they belong to are included in the result | ||
# keep 1 document from each group of duplcates and mark the others to remove | ||
# https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.duplicated.html | ||
docs_to_remove = duplicates.df.map_partitions( | ||
lambda x: x[x.group.duplicated(keep="first")] | ||
) | ||
|
||
# When there are few duplicates we can compute the results to a list and use `isin`. | ||
result = input_dataset.df[ | ||
~input_dataset.df[dataset_id_field].isin( | ||
docs_to_remove[dataset_id_field].compute() | ||
) | ||
] | ||
write_to_disk(result, output_dir, output_type=filetype) | ||
print(time.time() - t0) | ||
|
||
|
||
def attach_args( | ||
parser=argparse.ArgumentParser( | ||
formatter_class=argparse.ArgumentDefaultsHelpFormatter | ||
), | ||
): | ||
return add_distributed_args(parser) | ||
|
||
|
||
if __name__ == "__main__": | ||
main(attach_args().parse_args()) |
This file was deleted.
Oops, something went wrong.
This file was deleted.
Oops, something went wrong.
53 changes: 0 additions & 53 deletions
53
examples/gpu_deduplication_example/create-list-of-exact-duplicate-ids.sh
This file was deleted.
Oops, something went wrong.
66 changes: 0 additions & 66 deletions
66
examples/gpu_deduplication_example/create-list-of-fuzzy-duplicate-ids.sh
This file was deleted.
Oops, something went wrong.
This file was deleted.
Oops, something went wrong.
This file was deleted.
Oops, something went wrong.
Oops, something went wrong.