Dear YAIB team,
I am attempting to reproduce the YAIB benchmarks but I am encountering a strict character validation error when setting up Weights & Biases.
The Issue:
When running the following command:
wandb sweep --verbose experiments/benchmark_classification.yml
The system prompts for a W&B API key but rejects any input that is not exactly 40 characters long. Modern W&B API keys (formatted as wandb_v1_...) are significantly longer than this legacy limit.
Troubleshooting already attempted:
-
Exporting the key directly via environment variables (export WANDB_API_KEY=...).
-
Manually editing configuration files/.netrc.
-
The system continues to enforce the length constraint regardless of the input method.
Environment:
YAIB Branch: development (as seen in my local setup)
Expected Behavior:
The system should accept modern, variable-length W&B API keys.
Thank you for your help!
All the best,
Georgina Petropoulou
Dear YAIB team,
I am attempting to reproduce the YAIB benchmarks but I am encountering a strict character validation error when setting up Weights & Biases.
The Issue:
When running the following command:
wandb sweep --verbose experiments/benchmark_classification.ymlThe system prompts for a W&B API key but rejects any input that is not exactly 40 characters long. Modern W&B API keys (formatted as wandb_v1_...) are significantly longer than this legacy limit.
Troubleshooting already attempted:
Exporting the key directly via environment variables (export WANDB_API_KEY=...).
Manually editing configuration files/.netrc.
The system continues to enforce the length constraint regardless of the input method.
Environment:
OS: macOS Big Sur, version 11.7.10 (20G1427)
W&B Version: version 0.18.7
YAIB Branch: development (as seen in my local setup)
Expected Behavior:
The system should accept modern, variable-length W&B API keys.
Thank you for your help!
All the best,
Georgina Petropoulou