DGL errors with Jupyter Notebook in Sagemaker (FileNotFoundError: Cannot find DGL C++ graphbolt library at [/opt/conda/lib/python3.10/site-packages/dgl/graphb)

I am exploring using DGL in a project. I tried to run one of the example in this github notebook:

https://github.com/aws-samples/aws-healthcare-lifescience-ai-ml-sample-notebooks/blob/main/workshops/Molecular-property-prediction/hiv-inhibitor-prediction-dgl/molecule-hiv-inhibitor-prediction-sagemaker.ipynb

But getting errors when importing DGL. See the logs below. Any help in solving this issue is greatly appreciated.

create a jupyterstudio notebook and run the appropriate cells from link above. Since the requirements file has an year old versions I just replaced the the requirements.txt to pick the latest version.

Logs

FileNotFoundError Traceback (most recent call last) Cell In[1], line 7 4 import torch.nn as nn 5 import pandas as pd ----> 7 import dgl 9 from dgllife.model import load_pretrained 10 from dgllife.utils import smiles_to_bigraph, EarlyStopping, Meter, CanonicalAtomFeaturizer, CanonicalBondFeaturizer

File /opt/conda/lib/python3.10/site-packages/dgl/init.py:16 13 from .logging import enable_verbose_logging # usort: skip 14 from .backend import backend_name, load_backend # usort: skip ---> 16 from . import ( 17 container, 18 cuda, 19 dataloading, 20 function, 21 ops, 22 random, 23 sampling, 24 storages, 25 ) 26 from ._ffi.base import version, DGLError 27 from ._ffi.function import ( 28 extract_ext_funcs, 29 get_global_func, 30 list_global_func_names, 31 register_func, 32 )

File /opt/conda/lib/python3.10/site-packages/dgl/dataloading/init.py:13 11 if F.get_preferred_backend() == "pytorch": 12 from .spot_target import * ---> 13 from .dataloader import * 14 from .dist_dataloader import *

File /opt/conda/lib/python3.10/site-packages/dgl/dataloading/dataloader.py:27 25 from ..batch import batch as batch_graphs 26 from ..cuda import GPUCache ---> 27 from ..distributed import DistGraph 28 from ..frame import LazyFeature 29 from ..heterograph import DGLGraph

File /opt/conda/lib/python3.10/site-packages/dgl/distributed/init.py:5 3 from .dist_context import exit_client, initialize 4 from .dist_dataloader import DistDataLoader ----> 5 from .dist_graph import DistGraph, DistGraphServer, edge_split, node_split 6 from .dist_tensor import DistTensor 7 from .graph_partition_book import GraphPartitionBook, PartitionPolicy

File /opt/conda/lib/python3.10/site-packages/dgl/distributed/dist_graph.py:11 7 from collections.abc import MutableMapping 9 import numpy as np ---> 11 from .. import backend as F, graphbolt as gb, heterograph_index 12 from .._ffi.ndarray import empty_shared_mem 13 from ..base import ALL, DGLError, EID, ETYPE, is_all, NID

File /opt/conda/lib/python3.10/site-packages/dgl/graphbolt/init.py:55 51 except Exception: # pylint: disable=W0703 52 raise ImportError("Cannot load Graphbolt C++ library") ---> 55 load_graphbolt()

File /opt/conda/lib/python3.10/site-packages/dgl/graphbolt/init.py:45, in load_graphbolt() 43 path = os.path.join(dirname, "graphbolt", basename) 44 if not os.path.exists(path): ---> 45 raise FileNotFoundError( 46 f"Cannot find DGL C++ graphbolt library at {path}" 47 ) 49 try: 50 torch.classes.load_library(path)

FileNotFoundError: Cannot find DGL C++ graphbolt library at [/opt/conda/lib/python3.10/site-packages/dgl/graphb