Jeffrey Martin

Also on the question: > Should site config and core config support JSON format? I would say `site` config should support both, however `core` is not a user configurable file...

A proposal for this, needs thought and feedback. Scenarios needing support: * module level rename for all classes in the module, example `knownbadsignatures` -> `av_spam_scanning` from #850 * class level...

Working thru backward compatible config and had some thoughts to work out: * should the config just ETL and updated before executing the harness? * `plugin_spec`/`detector_spec`/`buff_spec` are logged in the...

Circling back here, given the number of changes requested could this be separated to isolate the various changes? I can see the following possible independent PRs: * enable `restGenerator` to...

I am looking into extracting the `verify_ssl` and `proxies` support this week while maintaining attribution to @au70ma70n as the original author. Once opened the PRs will reference this original work....

Closing for now, the proxy support and TLS/SSL validation suppression for `rest` generators landed independently. The token refresh looks interesting though likely needs some more work. Multi-endpoint patterns are still...

The issue was identified when attempting to validate [this linked comment](https://github.com/NVIDIA/garak/pull/1051#issuecomment-2665542380).

This is implemented in `OpenAICompatible` any `nim` class inherits it as long as the class provides a `context_len`, which can be set via config or a pattern similar to OpenAI...

Updates in 95452e0, create a consolidated method to support `max_tokens` based on an available `context_len` and shift `chat` clients to utilize `max_completion_tokens`. I suspect there may be some stated OpenAI...

Testing has indicated that many `nim` deployments will not accept `max_completion_tokens` more thought is needed on how to ensure the right key is submitted without forcing creation of esoteric configuration...