All pull requests

Pull requests

• fix(aim344): propagate trusted EFA assignments to health checks

  #1325·KeitaW opened Nov 2, 2026Collaborator·

  ·

• fix(healthcheck): validate explicitly assigned G7e EFA interfaces

  #1324·KeitaW opened Nov 2, 2026Collaborator·

  ·

• fix(aim344): honor allocated Slurm resources and management routes

  #1323·KeitaW opened Nov 2, 2026Collaborator·

  ·

• fix(aim347): separate replica internal vLLM port scan starts

  #1322·KeitaW opened Nov 1, 2026Collaborator·5 of 7 tasks·

  ·

• fix(aws-pcs): restore OpenZFS home through systemd on reboot

  #1321·KeitaW opened Nov 1, 2026Collaborator·

  ·

• feat(inference-gateway): add streaming checks to 0.test-gateway.sh

  #1320·ayahuja opened Oct 30, 2026·

  ·

• Measure SDC detection schedule costs end to end

  #1318·allela-roy opened Oct 29, 2026Contributor·8 tasks done·

  ·

• cosmos3: re-pin cosmos-framework to 1a2f03a (uv 0.11.29) + README DROID-dataset correction

  #1317·bluecrayon52 opened Oct 29, 2026Contributor·8 tasks done·

  ·

• feat(nccl-tests): Add an EFA latency matrix step for Amazon EKS

  #1313·rajjainl opened Oct 27, 2026Contributor·

  ·
  14

• feat(vllm): Add Kimi K2 across two instances (PP2xTP8, TP16)

  #1311·rajjainl opened Oct 27, 2026Contributor·

  ·
  12

• disaggregated-inference: choose the KV page size and engine environment variables in config.json

  #1305·littlemex opened Oct 27, 2026Contributor·7 of 8 tasks·

  ·
  1

• Inference Gateway IRSA Support

  #1293·mguaws opened Oct 19, 2026Contributor·11 tasks done·

  ·
  1

• feat(eks-observability): selectable scrape configs, SGLang/EFA metrics, and SGLang PD dashboard

  #1290·ybalbert001 opened Oct 15, 2026Contributor·

  ·
  11

• Add Ray on HyperPod SFT + LoRA example with Ray Serve

  #1277·varrajan opened Oct 8, 2026·4 of 5 tasks·

  ·
  35

• Add dense LLM training, checkpoint recovery and serving experiments

  #1271·KeitaW opened Sep 24, 2026Collaborator·2 of 3 tasks·

  ·

• Fix/healthcheck result severity clobber

  #1270·Zhenye-Na opened Sep 23, 2026Contributor·5 of 8 tasks·

  ·

• Riv2026/aim347 content

  #1268·KeitaW opened Sep 21, 2026Collaborator·

  ·

• Fix ramp/collector/transport evidence defects and record the measured paired result (AIM345)

  #1266·whn09 opened Sep 16, 2026Collaborator·

  ·
  1

• feat(megatron-bridge): add upstream dev DeepEP v2 EFA training configuration

  #1265·KeitaW opened Sep 16, 2026Collaborator·2 of 4 tasks·

  ·

• feat(deepep-efa): NVIDIA Dynamo DeepEP-V2 MoE all-to-all over EFA (NCCL-GIN CPU-proxy)

  #1256·dmvevents opened Sep 4, 2026Contributor·5 tasks done·

  ·
  75

• feat(deepep-efa): NeMo-RL GRPO + Megatron Shape-Y MoE all-to-all over EFA (DeepEP-V2 NCCL-GIN)

  #1242·dmvevents opened Aug 25, 2026Contributor·

  ·
  81

• feat(deepep-efa): TensorRT-LLM NcclEP MoE all-to-all over EFA (NCCL-GIN CPU-proxy)

  #1240·dmvevents opened Aug 25, 2026Contributor·

  ·
  105

• feat(verl): GRPO + Megatron/FSDP LoRA recipe on EKS with Hydra config and eval pipeline

  #1232·mvinci12 opened Aug 17, 2026Contributor·7 tasks done·

  ·
  68

• feat(deepep-efa): vLLM DeepEP-V2 MoE all-to-all over EFA via NCCL-GIN GDAKI (GPU-initiated)

  #1231·dmvevents opened Aug 14, 2026Contributor·

  ·
  8

• feat(deepep-efa): vLLM DeepEP-V2 MoE all-to-all over EFA (eager + non-eager)

  #1230·dmvevents opened Aug 14, 2026Contributor·

  ·
  171