Commits · 261881221886ea8abecd0767e73b2b95d5b15d3f · OpenDAS / dynamo · GitLab

20 Apr, 2026 4 commits
- feat(planner): own AIC interpolation; fix MoE-DEP bugs in rapid mode (#8335) · 26188122
  Hongkuan Zhou authored Apr 20, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
```
  26188122
- feat(planner/replay): KV reuse awareness in load + throughput scaling (#8314) · c388483a
  Hongkuan Zhou authored Apr 20, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
```
  c388483a
- fix(sglang): add enable_trace to diffusion worker ServerArgs stub (#8332) · 1d154dcb
  Krishnan Prashanth authored Apr 20, 2026
```
Signed-off-by: Krishnan Prashanth <kprashanth@nvidia.com>
```
  1d154dcb
- fix: propagate vLLM --stream-interval to Dynamo frontend (#8101) · 714b4234
  Roger Wang authored Apr 19, 2026
```
Signed-off-by: Roger Wang <hey@rogerw.io>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  714b4234
17 Apr, 2026 3 commits
- fix(replay/planner): planner-in-the-loop replay diagnostics + planner scaling logics fixes (#8280) · 0c7068dd
  Hongkuan Zhou authored Apr 17, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  0c7068dd
- fix: Only read prompt embeds from request if globally enabled (#8248) · 0d5bf022
  Graham King authored Apr 17, 2026
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  0d5bf022
- fix: Use vllm to load prompt embeds (#8228) · d644d88d
  Graham King authored Apr 17, 2026
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  d644d88d
16 Apr, 2026 2 commits
- feat: Expanding the hardware discovery to include RDMA, interconnect. (#7551) · 63787939
  devivasudevan authored Apr 16, 2026
  
  63787939
- fix(sglang): pass stop_token_ids to engine for proactive EOS detection (#8084) · 7d46806d
  Ryan McCormick authored Apr 15, 2026
  
  7d46806d
15 Apr, 2026 12 commits
- feat(planner): add advisory mode for scaling decisions (#8244) · 3da6f4d4
  Hongkuan Zhou authored Apr 15, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  3da6f4d4
- chore: flux benchmarking script + code clean (#8083) · 2d86b81d
  Ayush Agarwal authored Apr 15, 2026
```
Signed-off-by: ayushag <ayushag@nvidia.com>
```
  2d86b81d
- fix(trtllm): read runtime config from resolved engine args (#8217) · 2e7a1e6c
  Thomas Montfort authored Apr 15, 2026
```
Signed-off-by: Thomas Montfort <tmontfort@nvidia.com>
Signed-off-by: tmontfort <tmontfort@nvidia.com>
```
  2e7a1e6c
- feat(multimodal): forward mm_processor_kwargs through pipeline for use_audio_in_video (#8150) · e232bec0
  Neal Vaidya authored Apr 15, 2026
```
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  e232bec0
- chore: Remove unused vLLM multimodal chat processor (#8229) · 2ac22df7
  Graham King authored Apr 15, 2026
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  2ac22df7
- fix(replay): drain traffic metrics only on throughput ticks (#8232) · b92c9593
  Hongkuan Zhou authored Apr 15, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  b92c9593
- feat: add TRT-LLM sleep/wake integration with GMS (#7575) · d96a2cf1
  hhzhang16 authored Apr 15, 2026
```
Co-authored-by: Schwinn Saereesitthipitak <schwinns@nvidia.com>
Co-authored-by: Dmitry Tokarev <dtokarev@nvidia.com>
```
  d96a2cf1
- fix(sglang): always populate max_num_batched_tokens in MDC (#8220) · 119149f2
  ishandhanani authored Apr 15, 2026
  
  119149f2
- feat(replay): planner-in-the-loop offline replay (#8187) · 89d67172
  jthomson04 authored Apr 14, 2026
```
Signed-off-by: jthomson04 <jwillthomson19@gmail.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  89d67172
- feat(metrics): auto-inject worker_id label into metrics (#8089) · cabdaebd
  Thomas Montfort authored Apr 14, 2026
```
Signed-off-by: tmontfort <tmontfort@nvidia.com>
Signed-off-by: Thomas Montfort <tmontfort@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  cabdaebd
- fix(trtllm): prevent decode worker segfault on prefill scale-down (#7933) · b13a440e
  MatejKosec authored Apr 14, 2026
```
Signed-off-by: Matej Kosec <mkosec@nvidia.com>
```
  b13a440e
- docs(profiler): add replay_optimize sweep README (#8195) · 9a07ca15
  Yan Ru Pei authored Apr 14, 2026
```
Signed-off-by: PeaBrane <yanrpei@gmail.com>
```
  9a07ca15
14 Apr, 2026 11 commits
- feat: add unified backend architecture with DynamoBackend (#8003) · da40db40
  Tanmay Verma authored Apr 14, 2026
```
Signed-off-by: Tanmay Verma <tanmayv@nvidia.com>
```
  da40db40
- feat(planner): add live diagnostics dashboard HTTP endpoint (#8168) · 091cdb51
  Hongkuan Zhou authored Apr 14, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  091cdb51
- fix(planner): don't block agg decode scaling when max_num_batched_tokens is missing (#8196) · d185c881
  Hongkuan Zhou authored Apr 14, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  d185c881
- feat: remove enable_nats usage (#7265) · af0ff07c
  Biswa Panda authored Apr 14, 2026
  
  af0ff07c
- feat(lora): add LoRA support for SGLang (#4769) · f9839161
  Biswa Panda authored Apr 14, 2026
```
Signed-off-by: Keiven Chang <keivenchang@users.noreply.github.com>
```
  f9839161
- fix: Reject local file inputs in ImageLoader (#8158) · 56f02422
  Graham King authored Apr 14, 2026
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  56f02422
- feat: propagate otel tracing to trtllm E/P/D workers (#7592) · 34ce777d
  Jie Hao authored Apr 14, 2026
  
  34ce777d
- feat: expose multimodal embedding cache metrics via Prometheus (#8031) · 96476be5
  Krishnan Prashanth authored Apr 14, 2026
```
Signed-off-by: Krishnan Prashanth <kprashanth@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  96476be5
- fix(gms): restore default allocator cleanup before checkpoint (#8049) · 6f5b460c
  Schwinn Saereesitthipitak authored Apr 13, 2026
```
Signed-off-by: Schwinn Saereesitthipitak <schwinns@nvidia.com>
```
  6f5b460c
- fix(replay): short-circuit .npz planner imports (#8050) · 31909ca3
  Yan Ru Pei authored Apr 13, 2026
```
Signed-off-by: PeaBrane <yanrpei@gmail.com>
```
  31909ca3
- feat(planner): add optimization_target for easy-mode scaling (#8137) · 9cf9ef18
  Hongkuan Zhou authored Apr 13, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  9cf9ef18
13 Apr, 2026 6 commits
- feat(sglang): add ephemeral KV session routing (#7665) · 9498f016
  ishandhanani authored Apr 13, 2026
```
Signed-off-by: Ishan Dhanani <ishandhanani@gmail.com>
```
  9498f016
- fix: disagg_same_gpu - profiling markers, GPU pinning, and memory args (#7996) · e3e728a8
  Keiven C authored Apr 13, 2026
```
Signed-off-by: Keiven Chang <keivenchang@users.noreply.github.com>
```
  e3e728a8
- feat: adds support to pass local tokenizer path (#8128) · e41e1715
  yunzhoul-nv authored Apr 13, 2026
```
Signed-off-by: Yunzhou Liu <232973175+yunzhoul-nv@users.noreply.github.com>
```
  e41e1715
- docs(router): split router docs into focused pages (#8122) · 49087845
  Yan Ru Pei authored Apr 13, 2026
```
Signed-off-by: PeaBrane <yanrpei@gmail.com>
```
  49087845
- chore: upgrade sglang dependency from 0.5.9 to 0.5.10.post1 (#7997) · e48de6aa
  Krishnan Prashanth authored Apr 13, 2026
```
Signed-off-by: KrishnanPrash <140860868+KrishnanPrash@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
```
  e48de6aa
- feat: route requests by device type and load for sglang epd (#7215) · a7d51b39
  Wang, Yi authored Apr 14, 2026
```
Signed-off-by: Wang, Yi <yi.a.wang@intel.com>
```
  a7d51b39
11 Apr, 2026 2 commits
- feat(planner): add diagnostics metrics, rename to dynamo_planner_*, and add HTML report (#8078) · f8920708
  Hongkuan Zhou authored Apr 10, 2026
  
  f8920708
- feat: Request Migration disable on max seq len exceeded (#8020) · f6976e7f
  Jacky authored Apr 10, 2026
```
Signed-off-by: Jacky <18255193+kthui@users.noreply.github.com>
```
  f6976e7f