-
## Introduction
Transfer the safety committee glossary in google doc to the zephyr /doc to have it publicly available.
### Problem description
The safety committee glossary shall be publicly …
-
## Description
Users express the need for data schema evaluation to enable "fail-fast" capabilities during data loading and consistency checks before execution. They highlight the potential benefits …
-
Currently, build graph calculations and `BUILD` file evaluation are pinned to the `__local__` environment (see #17129), meaning that it will not run in a `docker`, or `remote` environment. This is acc…
-
Here are some ideas and potential areas of research for Tensort:
- Model analysis and interpretability: Develop new techniques for analyzing and understanding what large language models have learned …
-
# Main todos:
- [ ] Check whether all move sorting table are correctly allocated, cleared or scaled, prepare some tests for them. Maybe clearing is not necessary
- [ ] Prepare and test better coef …
-
This code won't compile, for nullable `y`:
```dart
int x = /* ... */;
int? y = /* ... */;
if (y != null && x > y) { /* ... */ }
```
What is needed is
```dart
int x = /* ... */;
int? y =…
-
The Safety Committee / WG started the effort to gather software requirements for the Zephyr project which are needed for the safety efforts and the quality of the project.
For the management of the r…
-
For completeness and to address one of the reviewers comments, we should also train RL, including the safety filter.
@Erfi: Where in the current setup can I add the safety filter (which is kind of …
-
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? And how did …
-
The current evaluation metrics supported by `llm-eval` are robust. However, upon reviewing the documentation, I found that the current repo doesn't account for evaluating model toxicity. Assessing LLM…