Tony Au-Yeung

fix: oras-go client should fallback to docker config if no credentials specified

1

I'm using https://github.com/argoproj/argo-cd/issues/17279 to authenticate to Google Artifact Registry as a helm registry using workload identity. Unlike https://github.com/argoproj/argo-cd/issues/10218, the other solution does not require installing ESO. This works in general...

Fixes https://github.com/elevenlabs/elevenlabs-python/issues/366 The issue is because the server now supports gzip responses and the client is allowing gzip as an encoding. This is a problem because people might want to...

[Feature] Prompt lookup speculative decoding for LLM API

4

It looks like the model runner API supports prompt lookup speculative decoding: https://github.com/NVIDIA/TensorRT-LLM/tree/main/examples/prompt_lookup However, it doesn't seem to be part of the LLM API yet: https://github.com/NVIDIA/TensorRT-LLM/blob/3ee4332fb183bf09a8a8a577bb3dd9a8e68f29f6/tensorrt_llm/llmapi/llm_args.py#L851-L854

Please see the FAQ in our main README.md before submitting your issue. ### Use case(s) - what problem will this feature solve? If a unary request takes a long time,...

sentinel client should listen for switch-master

1

After a failover, we rely on a retry to pick up the new master. It seems like other client libraries will listen for the switch-master and automatically point to it....