uber/neuropod

ValueError: Expected requirements of the form name==version but got --index-url

qiyanz opened this issue · 2 comments

qiyanz commented

Bug

Deeplearning Inference could not pass the local test for models and got the following error

LoadNeuropod modelPath = ./example/torch_addition/neuropod failed: ValueError: Expected requirements of the form name==version but got --index-url

To Reproduce

Steps to reproduce the behavior:

  1. Run the command that tests c_model in the terminal with neuropod 0.3.0 rc

Expected behavior

All tests could pass

Environment

  • Neuropod Version: 0.3.0rc
  • OS:macOS
  • Language: Python
  • Python version: 2.7
  • Using OPE: no

If this bug report is about running a specific model:

  • Neuropod backend (e.g. TensorFlow, PyTorch, Keras, TorchScript, Python): PyTorch
  • Framework version (e.g. 1.3.0): 1.4

If running on GPU: No

  • CUDA/cuDNN version:
  • The output of nvidia-smi:
  • Any other relevant information:

Additional context

VivekPanyam commented

As we talked about offline, the solution will probably be to add --no-emit-index-url and possibly other flags to

neuropod/source/python/neuropod/utils/pip_utils.py

Lines 24 to 37 in ca7e3b8

subprocess.check_call(
[
sys.executable,
"-m",
"piptools",
"compile",
"--no-header",
"--no-annotate",
"--allow-unsafe",
"-q",
"-o",
lockfile,
requirements,
],

VivekPanyam commented

Closed by #480. Thanks for the PR!