Labels

• api-breaking

  Accepted LLM API contract change that is backwards-incompatible
• api-compatible

  Accepted LLM API contract change that is backwards-compatible
• AutoDeploy

  <NV> AutoDeploy Backend
• AutoDeploy/Dashboard

  <NV> AutoDeploy Backend: Dashboard related
• AutoDeploy/llmc-blocker

  <NV> Tag for issues that are blocking AutoDeploy standalone repo
• breaking-api-change

  When the PR breaks backwards compatibility
• bug

  Something isn't working
• Cherry-pick

  It's a label that applies to Cherry-pick PR.
• Community Engagement

  help/insights needed from community
• Community want to contribute

  PRs initiated from Community
• CUDA Graph
• Customized kernels

  <NV>Specialized/modified CUDA kernels in TRTLLM for LLM ops, beyond standard TRT. Dev & perf.
• Decoding/Sampling

  <NV>Token sampling algorithms in TRTLLM for text gen (top-k, top-p, beam).
• deepseek-v4
• dependencies

  Pull requests that update a dependency file
• Disaggregated serving

  <NV>Deploying with separated, distributed components (params, kv-cache, compute). Arch & perf.
• Doc

  <NV>TRTLLM's textual/illustrative materials: API refs, guides, tutorials. Improvement & clarity.
• duplicate

  This issue or pull request already exists
• Ease of Use

  Items about improving or complaints about TRTLLM ease of use
• feature request

  New feature or request. This includes new model, dtype, functionality support
• Frontend

  <NV>Frontend of the LLM workflow
• functionality issue
• gagam-slack-watch

  Label for getting slack notifications via GitHub Slack app
• General perf

  <NV>Broad performance issues not specific to a particular component
• help wanted

  Extra attention is needed
• Inference runtime

  <NV>General operational aspects of TRTLLM execution not in other categories.
• Infra

  <NV>automated tests, build checks, github actions, system stability & efficiency.
• Installation

  Setting up and building TRTLLM: compilation, pip install, dependencies, env config, CMake.
• Investigating
• KV-Cache Management

  kv-cache management for efficient LLM inference