Search results

[CI] Add GLM 4.7 to Megatron Models CI

Status: Open.

#1681 In NovaSky-AI/SkyRL;

· SumanthRH opened on May 17, 2026

WorkerWrap.load_weights silently corrupts unquantized MoE rollouts after the first weight sync since vllm>=0.20.0

Status: Open.

#1680 In NovaSky-AI/SkyRL;

· jamesbraza opened on May 17, 2026

create_ray_wrapped_inference_engines drops the engine-core child's root cause on init failure

Status: Open.

#1673 In NovaSky-AI/SkyRL;

· jamesbraza opened on May 15, 2026

RemoteInferenceEngine swallows vLLM's 404, surfaces as opaque IndexError

Status: Open.

#1672 In NovaSky-AI/SkyRL;

· jamesbraza opened on May 15, 2026

Fix metrics assertions for Tinker-API based E2E CI runs

Status: Open.

#1664 In NovaSky-AI/SkyRL;

· SumanthRH opened on May 14, 2026

[cleanup] remove flashrl mentions from the repo

Status: Open.

#1658 In NovaSky-AI/SkyRL;

· erictang000 opened on May 13, 2026

[tinker] SkyRLTrainBackend tinker multi-tenancy improvements

Status: Open.

#1654 In NovaSky-AI/SkyRL;

· erictang000 opened on May 12, 2026

[vllm] test out ray v1 executor in vLLM

Status: Open.

#1653 In NovaSky-AI/SkyRL;

· erictang000 opened on May 12, 2026

[tinker] Enable multi-lora training for async RL

Status: Open.

#1647 In NovaSky-AI/SkyRL;

· SumanthRH opened on May 11, 2026

[tinker] Support additional Tinker LoraConfig parameters

Status: Open.

#1632 In NovaSky-AI/SkyRL;

· erictang000 opened on May 7, 2026

[vllm] Investigate whether NCCL_CUMEM_ENABLE should be set to 0

Status: Open.

#1630 In NovaSky-AI/SkyRL;

· erictang000 opened on May 7, 2026

[models] Qwen3.5 Issue Tracker

Status: Open.

#1626 In NovaSky-AI/SkyRL;

· erictang000 opened on May 6, 2026