google-research / google-research/circuit_training

Issues with TensorFlow and reverb compatibility in AlphaChip setup

Open
#71 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
1.7k
Forks
273
PR merge metrics
No merged PRs in 30d

Description

I've been attempting to set up AlphaChip for my project but encountered several errors related to TensorFlow and reverb compatibility. Here's a detailed summary of the issue:

Problem Overview:
Error Messages: Undefined symbols in libpybind.so and version mismatches between TensorFlow and reverb.

Environment: Python 3.10.10, TensorFlow 2.18.0, dm-reverb 0.10.0, CUDA 12.1, cuDNN 8.9.

Steps Taken:
Installed Dependencies:

TensorFlow 2.18.0

dm-reverb 0.10.0

Set environment variables for DREAMPlace and placement cost binary.

Checked CUDA and cuDNN Compatibility:

Confirmed CUDA 12.1 and cuDNN 8.9 are installed.

Ran scripts to check if TensorFlow recognizes the GPU.

Investigated Protobuf Version:

Checked Protobuf version to ensure compatibility with TensorFlow.

Created a Clean Environment:

Set up a new conda environment with Python 3.10.

Reinstalled TensorFlow and reverb in the new environment.

Despite these efforts, the issue persists. I would appreciate any guidance or insights into resolving these compatibility issues. If possible, please provide suggestions for troubleshooting or point me towards resources that could help.

Contributor guide

Open the contributing guide

Research direction

No source files or tests are identified. Start by reproducing the AlphaChip setup with Python 3.10.10, TensorFlow 2.18.0, dm-reverb 0.10.0, CUDA 12.1, and cuDNN 8.9, then isolate the undefined-symbol and version-mismatch errors. Done means identifying compatible dependency versions or documenting a confirmed resolution.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, tensorflow
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.