NVIDIA · sarahyurick · Nov 5, 2024 · Nov 5, 2024 · Nov 5, 2024
diff --git a/tutorials/image-curation/image-curation.ipynb b/tutorials/image-curation/image-curation.ipynb
@@ -51,8 +51,7 @@
    "source": [
     "!pip install cython ipywidgets aiofiles\n",
     "# Install from source by default\n",
-    "!pip install --extra-index-url https://pypi.nvidia.com ../../[image]\n",
-    "%env DASK_DATAFRAME__QUERY_PLANNING False"
+    "!pip install --extra-index-url https://pypi.nvidia.com ../../[image]"
    ]
   },
   {

diff --git a/tutorials/pretraining-data-curation/start-distributed-notebook.sh b/tutorials/pretraining-data-curation/start-distributed-notebook.sh
@@ -62,7 +62,6 @@ export CUDF_SPILL="1"
 export RMM_SCHEDULER_POOL_SIZE="1GB"
 export RMM_WORKER_POOL_SIZE="72GiB"
 export LIBCUDF_CUFILE_POLICY=OFF
-export DASK_DATAFRAME__QUERY_PLANNING=False
 
 
 # =================================================================

diff --git a/tutorials/pretraining-vietnamese-data-curation/pretraining-vietnamese-data-curation.ipynb b/tutorials/pretraining-vietnamese-data-curation/pretraining-vietnamese-data-curation.ipynb
@@ -131,11 +131,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
-    "import nemo_curator\n",
     "from dask.distributed import Client, LocalCluster\n",
     "\n",
     "# Start a Dask cluster with 12 workers, each limited at 64GB of memory. \n",
@@ -708,17 +707,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import os\n",
-    "os.environ[\"DASK_DATAFRAME__QUERY_PLANNING\"] = \"False\""
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 12,
+   "execution_count": null,
    "metadata": {},
    "outputs": [
     {
@@ -734,7 +723,6 @@
    ],
    "source": [
     "from nemo_curator.utils.distributed_utils import get_client\n",
-    "import dask.dataframe\n",
     "\n",
     "def pre_imports():\n",
     "    import cudf \n",
@@ -1030,13 +1018,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
     "import os\n",
-    "from datasets import load_dataset as load_hf_dataset\n",
-    "from datasets import DownloadConfig "
+    "from datasets import load_dataset as load_hf_dataset"
    ]
   },
   {
@@ -1133,14 +1120,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
     "from nemo_curator import Modify\n",
-    "from nemo_curator.modifiers import UnicodeReformatter\n",
-    "from nemo_curator.utils.distributed_utils import read_data, write_to_disk\n",
-    "from nemo_curator.utils.file_utils import get_all_files_paths_under\n",
+    "from nemo_curator.utils.distributed_utils import write_to_disk\n",
     "from nemo_curator.datasets import DocumentDataset"
    ]
   },

diff --git a/tutorials/zyda2-tutorial/0_processing/process_dclm.py b/tutorials/zyda2-tutorial/0_processing/process_dclm.py
@@ -1,8 +1,5 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 from dask.distributed import Client, LocalCluster
 from helper import process_data

diff --git a/tutorials/zyda2-tutorial/0_processing/process_dolma_cc.py b/tutorials/zyda2-tutorial/0_processing/process_dolma_cc.py
@@ -1,8 +1,5 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 from dask.distributed import Client, LocalCluster
 from helper import process_data

diff --git a/tutorials/zyda2-tutorial/0_processing/process_fwe2.py b/tutorials/zyda2-tutorial/0_processing/process_fwe2.py
@@ -1,10 +1,7 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import ctypes
 import gc
 import logging
+import os
 from pathlib import Path
 
 from dask.distributed import Client, LocalCluster

diff --git a/tutorials/zyda2-tutorial/0_processing/process_zyda.py b/tutorials/zyda2-tutorial/0_processing/process_zyda.py
@@ -1,8 +1,5 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 from dask.distributed import Client, LocalCluster
 from helper import process_data

diff --git a/tutorials/zyda2-tutorial/1_fuzzy_dedup/0_minhash.py b/tutorials/zyda2-tutorial/1_fuzzy_dedup/0_minhash.py
@@ -1,8 +1,5 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 import time
 
 import dask_cudf

diff --git a/tutorials/zyda2-tutorial/1_fuzzy_dedup/1_lsh.py b/tutorials/zyda2-tutorial/1_fuzzy_dedup/1_lsh.py
@@ -1,8 +1,5 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 import time
 
 import cudf

diff --git a/tutorials/zyda2-tutorial/1_fuzzy_dedup/2_buckets_to_edges.py b/tutorials/zyda2-tutorial/1_fuzzy_dedup/2_buckets_to_edges.py
@@ -1,8 +1,5 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 import time
 
 import dask_cudf

diff --git a/tutorials/zyda2-tutorial/1_fuzzy_dedup/3_connected_components.py b/tutorials/zyda2-tutorial/1_fuzzy_dedup/3_connected_components.py
@@ -1,8 +1,5 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 import time
 
 from nemo_curator.modules.fuzzy_dedup import ConnectedComponents

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/0_id_mapping.py b/tutorials/zyda2-tutorial/2_dupes_removal/0_id_mapping.py
@@ -1,9 +1,6 @@
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import json
 import logging
+import os
 
 import cudf
 import dask_cudf

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/1_id_conversion.py b/tutorials/zyda2-tutorial/2_dupes_removal/1_id_conversion.py
@@ -1,9 +1,6 @@
 import json
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 import dask.dataframe as dd
 from dask.distributed import Client, LocalCluster

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/2_compute_counts.py b/tutorials/zyda2-tutorial/2_dupes_removal/2_compute_counts.py
@@ -1,11 +1,8 @@
 import json
+import logging
 import os
 import time
 
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
-import logging
-
 import dask.dataframe as dd
 import pandas as pd
 from dask.distributed import Client, LocalCluster

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/3_prep_dupes.py b/tutorials/zyda2-tutorial/2_dupes_removal/3_prep_dupes.py
@@ -1,9 +1,6 @@
 import json
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 import dask.dataframe as dd
 import pandas as pd

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/4_get_dupes_dclm.py b/tutorials/zyda2-tutorial/2_dupes_removal/4_get_dupes_dclm.py
@@ -1,9 +1,6 @@
 import json
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 import dask.dataframe as dd
 from dask.distributed import Client, LocalCluster

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/4_get_dupes_dolma-cc.py b/tutorials/zyda2-tutorial/2_dupes_removal/4_get_dupes_dolma-cc.py
@@ -1,9 +1,6 @@
 import json
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 import dask.dataframe as dd
 from dask.distributed import Client, LocalCluster

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/5_get_dupes_zyda.py b/tutorials/zyda2-tutorial/2_dupes_removal/5_get_dupes_zyda.py
@@ -1,9 +1,6 @@
 import json
-import os
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import os
 
 import dask.dataframe as dd
 from dask.distributed import Client, LocalCluster

diff --git a/tutorials/zyda2-tutorial/2_dupes_removal/remove_dupes.py b/tutorials/zyda2-tutorial/2_dupes_removal/remove_dupes.py
@@ -1,12 +1,9 @@
 import argparse
 import json
+import logging
 import os
 import time
 
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
-import logging
-
 import dask.dataframe as dd
 from dask.distributed import Client, LocalCluster
 

diff --git a/tutorials/zyda2-tutorial/3_quality_model/run_quality_classifier.py b/tutorials/zyda2-tutorial/3_quality_model/run_quality_classifier.py
@@ -1,11 +1,8 @@
 import argparse
+import logging
 import os
 import time
 
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
-import logging
-
 from nemo_curator.classifiers import QualityClassifier
 from nemo_curator.datasets import DocumentDataset
 from nemo_curator.utils.distributed_utils import get_client, get_num_workers

diff --git a/tutorials/zyda2-tutorial/4_filtering/filter_fwe.py b/tutorials/zyda2-tutorial/4_filtering/filter_fwe.py
@@ -1,10 +1,7 @@
+import logging
 import os
 import time
 
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
-import logging
-
 import dask.dataframe as dd
 import pyarrow as pa
 from dask.distributed import Client, LocalCluster

diff --git a/tutorials/zyda2-tutorial/4_filtering/filter_quality.py b/tutorials/zyda2-tutorial/4_filtering/filter_quality.py
@@ -1,10 +1,6 @@
 import argparse
-import os
-import time
-
-os.environ["DASK_DATAFRAME__QUERY_PLANNING"] = "False"
-
 import logging
+import time
 
 import dask.dataframe as dd
 import pyarrow as pa