Spacy not working for GPU on cuda 117

[batman202012]

How to reproduce the behaviour

Errors out on line 98 of the attached code (it is a .py but github won't let me attach unless it's a .txt) with traceback:

comparisons.txt

traceback.txt

Your Environment

• Operating System:
• Python Version Used: 3.9.13
• spaCy Version Used: 3.4.1 (with cuda 117 enable)
• Environment Information:
  The other pip packages in the environment are listed in pip_packages.txt attached:

pip_packages.txt

My nvcc verison is:
Build cuda_11.7.r11.7/compiler.31294372_0

My cuda verison is: 11.7

My nvidia driver version is: 516.59

My CUDNN version is: v 8.4.1.50

Running on NVIDIA rtx 2060 it is the only card attached

The only reason I have torch for cuda 11.6 is because there is no pip version for cuda 11.7 available and I was unable to build it from source

[adrianeboyd]

This looks more like it might be a problem with CUDA or torch rather than spacy itself. The attached file of pip packages was empty, can you provide the output of pip freeze?

You can also test a simpler example to see if it works or provides a more informative error message:

import spacy
spacy.require_gpu()
nlp = spacy.load("en_core_web_trf")
doc = nlp("This is a test sentence.")

[batman202012]

Looks like it works here but not in my code? it returns Traceback (most recent call last):

File "C:\Users\jnjbr\redenv\lib\site-packages\discord\client.py", line 343, in _run_event
   await coro(*args, **kwargs)
 File "C:\Users\jnjbr\Desktop\test_bot\cogs\CogManager\cogs\chatter\chat.py", line 716, in on_message_without_command
   future = await self.loop.run_in_executor(
 File "C:\Users\jnjbr\AppData\Local\Programs\Python\Python39\lib\concurrent\futures\thread.py", line 58, in run
   result = self.fn(*self.args, **self.kwargs)
 File "C:\Users\jnjbr\Desktop\test_bot\cogs\Downloader\lib\chatterbot\chatterbot.py", line 152, in generate_response
   output = adapter.process(input_statement, additional_response_selection_parameters)
 File "C:\Users\jnjbr\Desktop\test_bot\cogs\Downloader\lib\chatterbot\logic\best_match.py", line 32, in process
   for result in search_results:
 File "C:\Users\jnjbr\Desktop\test_bot\cogs\Downloader\lib\chatterbot\search.py", line 74, in search
   confidence = self.compare_statements(input_statement, statement)
 File "C:\Users\jnjbr\Desktop\test_bot\cogs\Downloader\lib\chatterbot\comparisons.py", line 15, in __call__
   return self.compare(statement_a, statement_b)
 File "C:\Users\jnjbr\Desktop\test_bot\cogs\Downloader\lib\chatterbot\comparisons.py", line 86, in compare
   document_b = self.nlp(statement_b.text)
 File "C:\Users\jnjbr\redenv\lib\site-packages\spacy\language.py", line 1025, in __call__
   error_handler(name, proc, [doc], e)
 File "C:\Users\jnjbr\redenv\lib\site-packages\spacy\util.py", line 1658, in raise_error
   raise e
 File "C:\Users\jnjbr\redenv\lib\site-packages\spacy\language.py", line 1020, in __call__
   doc = proc(doc, **component_cfg.get(name, {}))  # type: ignore[call-arg]
 File "spacy\pipeline\trainable_pipe.pyx", line 56, in spacy.pipeline.trainable_pipe.TrainablePipe.__call__
 File "C:\Users\jnjbr\redenv\lib\site-packages\spacy\util.py", line 1658, in raise_error
   raise e
 File "spacy\pipeline\trainable_pipe.pyx", line 52, in spacy.pipeline.trainable_pipe.TrainablePipe.__call__
 File "spacy\pipeline\tagger.pyx", line 143, in spacy.pipeline.tagger.Tagger.predict
 File "C:\Users\jnjbr\redenv\lib\site-packages\thinc\model.py", line 315, in predict
   return self._func(self, X, is_train=False)[0]
 File "C:\Users\jnjbr\redenv\lib\site-packages\thinc\layers\chain.py", line 55, in forward
   Y, inc_layer_grad = layer(X, is_train=is_train)
 File "C:\Users\jnjbr\redenv\lib\site-packages\thinc\model.py", line 291, in __call__
   return self._func(self, X, is_train=is_train)
 File "C:\Users\jnjbr\redenv\lib\site-packages\thinc\layers\chain.py", line 55, in forward
   Y, inc_layer_grad = layer(X, is_train=is_train)
 File "C:\Users\jnjbr\redenv\lib\site-packages\thinc\model.py", line 291, in __call__
   return self._func(self, X, is_train=is_train)
 File "C:\Users\jnjbr\redenv\lib\site-packages\spacy_transformers\layers\trfs2arrays.py", line 31, in forward
   dst, get_d_src = apply_alignment(model.ops, trf_data.align, src)
 File "C:\Users\jnjbr\redenv\lib\site-packages\spacy_transformers\align.py", line 45, in apply_alignment
   Y = Ragged(X[indices], cast(Ints1d, ops.asarray(align.lengths)))
 File "cupy\_core\core.pyx", line 1437, in cupy._core.core.ndarray.__getitem__
 File "cupy\_core\_routines_indexing.pyx", line 43, in cupy._core._routines_indexing._ndarray_getitem
 File "cupy\_core\core.pyx", line 782, in cupy._core.core.ndarray.take
 File "cupy\_core\_routines_indexing.pyx", line 144, in cupy._core._routines_indexing._ndarray_take
 File "cupy\_core\_routines_indexing.pyx", line 844, in cupy._core._routines_indexing._take
 File "cupy\_core\_kernel.pyx", line 849, in cupy._core._kernel.ElementwiseKernel.__call__
 File "cupy\_core\_kernel.pyx", line 138, in cupy._core._kernel._preprocess_args
 File "cupy\_core\_kernel.pyx", line 109, in cupy._core._kernel._preprocess_arg
 File "cupy\_core\_kernel.pyx", line 96, in cupy._core._kernel._check_peer_access
 File "cupy\cuda\device.pyx", line 98, in cupy.cuda.device._enable_peer_access
 File "cupy_backends\cuda\api\runtime.pyx", line 375, in cupy_backends.cuda.api.runtime.deviceCanAccessPeer
 File "cupy_backends\cuda\api\runtime.pyx", line 142, in cupy_backends.cuda.api.runtime.check_status
cupy_backends.cuda.api.runtime.CUDARuntimeError: cudaErrorInvalidDevice: invalid device ordinal

[adrianeboyd]

From the output above it looks like spacy itself is working fine in your venv, but something is going wrong for the CUDA or torch settings in your code?

[batman202012]

Could it possibly be an issue in the config.cfg for spacy mayhaps?

[batman202012]

Currently it is
default_config.txt

[batman202012]

Pretty sure I found my issue, the packages being used were not the ones being used in the environment. :( sorry to bother you guys