Robert Kirk

Model Parallelism and accelerate's usage of DDP aren't compatible

18

### System Info ```Shell - `Accelerate` version: 0.18.0 - Platform: Linux-5.4.0-124-generic-x86_64-with-glibc2.31 - Python version: 3.9.12 - PyTorch version (GPU?): 1.12.0 (True) - `Accelerate` default config: - compute_environment: LOCAL_MACHINE - distributed_type:...

This commit adds periodic logging of evaluation scores for the policy being trained. It also adds `num_levels` and `start_level` to the arguments. Based on code from @rraileanu

Hi there. I'm trying to run your code, and using the conda environment files provided to set up the environment. However, in both files, the requirements are incompatible. `gym==0.15.4` and...

Prompt and seed instructions for malicious instruction generation

1

In the paper in appendix B.2, you briefly describe how you generate the malicious instructions dataset. Could you share the prompt and seed instructions you used to generate this dataset?...