Closed Sparrow0hawk closed 4 years ago
Hey @Sparrow0hawk , I am reproducing the same error right now and looking for a fix. Thanks for bringing it to our attention!
I had the same issue
I installed numba again with conda install numba
with this output
Solving environment: done
## Package Plan ##
environment location: /usr/local
added / updated specs:
- numba
The following packages will be downloaded:
package | build
---------------------------|-----------------
certifi-2020.4.5.1 | py36_0 159 KB
openssl-1.1.1g | h7b6447c_0 3.8 MB
ca-certificates-2020.1.1 | 0 132 KB
conda-4.8.3 | py36_0 3.0 MB
llvmlite-0.32.1 | py36hd408876_0 17.6 MB
tqdm-4.46.0 | py_0 60 KB
conda-package-handling-1.6.1| py36h7b6447c_0 886 KB
numba-0.49.1 | py36h0573a6f_0 3.5 MB
------------------------------------------------------------
Total: 29.2 MB
The following NEW packages will be INSTALLED:
conda-package-handling: 1.6.1-py36h7b6447c_0
tqdm: 4.46.0-py_0
The following packages will be UPDATED:
certifi: 2020.4.5.1-py36h9f0ad1d_0 conda-forge --> 2020.4.5.1-py36_0
conda: 4.5.4-py36_0 --> 4.8.3-py36_0
llvmlite: 0.32.0-py36hfa65bc7_0 conda-forge --> 0.32.1-py36hd408876_0
numba: 0.49.1-py36h830a2c2_0 conda-forge --> 0.49.1-py36h0573a6f_0
openssl: 1.1.1g-h516909a_0 conda-forge --> 1.1.1g-h7b6447c_0
The following packages will be DOWNGRADED:
ca-certificates: 2020.4.5.1-hecc5488_0 conda-forge --> 2020.1.1-0
Proceed ([y]/n)? y
Downloading and Extracting Packages
certifi-2020.4.5.1 | 159 KB | : 100% 1.0/1 [00:00<00:00, 15.01it/s]
openssl-1.1.1g | 3.8 MB | : 100% 1.0/1 [00:00<00:00, 1.63it/s]
ca-certificates-2020 | 132 KB | : 100% 1.0/1 [00:00<00:00, 27.03it/s]
conda-4.8.3 | 3.0 MB | : 100% 1.0/1 [00:00<00:00, 1.55it/s]
llvmlite-0.32.1 | 17.6 MB | : 100% 1.0/1 [00:03<00:00, 3.03s/it]
tqdm-4.46.0 | 60 KB | : 100% 1.0/1 [00:00<00:00, 23.42it/s]
conda-package-handli | 886 KB | : 100% 1.0/1 [00:00<00:00, 6.71it/s]
numba-0.49.1 | 3.5 MB | : 100% 1.0/1 [00:01<00:00, 1.10s/it]
Preparing transaction: done
Verifying transaction: done
Executing transaction: done
but now I ran same demo code as you did. but it now gives new error
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-6-a95ca25217db> in <module>()
----> 1 import cudf
2 import io, requests
3
4 # download CSV file from GitHub
5 url="https://github.com/plotly/datasets/raw/master/tips.csv"
2 frames
/usr/local/lib/python3.6/site-packages/cudf/core/dataframe.py in <module>()
23
24 import cudf
---> 25 import cudf._lib as libcudf
26 from cudf.core import column
27 from cudf.core._sort import get_sorted_inds
AttributeError: module 'cudf' has no attribute '_lib'
when I run this code :
import cuml
# Create and populate a GPU DataFrame
df_float = cudf.DataFrame()
df_float['0'] = [1.0, 2.0, 5.0]
df_float['1'] = [4.0, 2.0, 1.0]
df_float['2'] = [4.0, 2.0, 1.0]
# Setup and fit clusters
dbscan_float = cuml.DBSCAN(eps=1.0, min_samples=1)
dbscan_float.fit(df_float)
print(dbscan_float.labels_)
It gives following error
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-8-435ab46c02e5> in <module>()
----> 1 import cuml
2
3 # Create and populate a GPU DataFrame
4 df_float = cudf.DataFrame()
5 df_float['0'] = [1.0, 2.0, 5.0]
/usr/local/lib/python3.6/site-packages/cuml/__init__.py in <module>()
17 from cuml.common.base import Base
18 from cuml.common.handle import Handle
---> 19 import cuml.common.cuda as cuda
20
21 from cuml.cluster.dbscan import DBSCAN
AttributeError: module 'cuml' has no attribute 'common'
Hey @Ahwar and @Sparrow0hawk and others, I pushed a PR yesterday that fixes this issue. Thanks for letting us know. It ended up that Numba, between 0.48 to 0.49, was being installed and it is apparently incompatible with RAPIDS 0.12 and 0.13. It works with 0.14. Props to Keith, a member of the team, who figured it out.
I'm trying to get RAPIDS working on google colabs but on import cudf i get an import error.
Steps to reproduce
Set up
Start colab session.
Connect to GPU runtime, in this instance I was allocated Tesla K80.
Install rapids
All looks good however...
Any recommendations?