Search results

Benchmark and support KV cache-aware routing for multi-replica deployments

Status: Open.

#16 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Evaluate and benchmark vLLM KV cache offloading options

Status: Open.

#15 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Implement autoscaling for model replicas based on traffic patterns

Status: Open.

#14 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Allow users to configure vLLM runtime parameters

Status: Open.

#13 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Add support for configurable routers and gateways

Status: Open.

#12 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Add llm-d as deployment target alongside KServe/vLLM

Status: Open.

#11 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Evaluate and potentially upgrade intent extraction LLM

Status: Open.

#10 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Design and document conversational interface UX flow

Status: Open.

#9 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Add GPU memory requirements to model catalog

Status: Open.

#8 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Integrate model accuracy benchmarks for task and domain suitability

Status: Open.

#7 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Research methods to reduce benchmark matrix size through extrapolation

Status: Open.

#6 In redhat-et/compass;

· anfredette opened on Nov 5, 2025

Replace synthetic benchmarks with real GuideLLM benchmark data

Status: Open.

#5 In redhat-et/compass;

· anfredette opened on Nov 5, 2025