[04:05:32] 06Machine-Learning-Team (Q2 FY2026-27): Scale TTS generation services to finish a batch run within a day - https://phabricator.wikimedia.org/T441983#12506999 (10kevinbazira) We ran a full batch-generation over the 4,500 article corpus (every section of every article) to replace the earlier estimate with a measur... [04:20:49] 06Machine-Learning-Team (Q2 FY2026-27): Scale TTS generation services to finish a batch run within a day - https://phabricator.wikimedia.org/T441983#12507025 (10kevinbazira) [10:05:43] 06Machine-Learning-Team (Q2 FY2026-27), 13Patch-For-Review: Create a namespace on ml-serve-eqiad for batch LLM workloads (llm-batch) - https://phabricator.wikimedia.org/T442150#12507563 (10OKarakaya-WMF) page updated: https://wikitech.wikimedia.org/wiki/Machine_Learning/LiftWing/Large_Language_Models @kevinba... [10:05:55] 06Machine-Learning-Team (Q2 FY2026-27), 13Patch-For-Review: Create a namespace on ml-serve-eqiad for batch LLM workloads (llm-batch) - https://phabricator.wikimedia.org/T442150#12507564 (10OKarakaya-WMF) [10:25:36] 06Machine-Learning-Team (Q2 FY2026-27), 13Patch-For-Review: Create a namespace on ml-serve-eqiad for batch LLM workloads (llm-batch) - https://phabricator.wikimedia.org/T442150#12507702 (10kevinbazira) >>! In T442150#12507562, @OKarakaya-WMF wrote: > page updated: https://wikitech.wikimedia.org/wiki/Machine_Le... [10:40:54] 06Machine-Learning-Team (Q2 FY2026-27), 06Data-Platform-SRE (2026-10-30 - 2026-11-20), 07Kubernetes: Extend KServe's LLMInferenceService by Gateway API Inference Extension and KEDA autoscaling - https://phabricator.wikimedia.org/T436650#12507795 (10DPogorzelski-WMF) Endpoint picker routing works end to end o... [10:53:54] 10Lift-Wing, 06Machine-Learning-Team (Q2 FY2026-27): Upgrade vLLM in AMD ROCm base image to v0.32.0 - https://phabricator.wikimedia.org/T442138#12507871 (10BWojtowicz-WMF) The image is published as `ml/amd-vllm032:gfx90agfx942rocm7.2.3pytorch2.13.0flash-attn2.8.3aiter0.1.24.post1vllm0.32.0-1`. I tested it wit... [11:09:04] 06Machine-Learning-Team (Q2 FY2026-27), 13Patch-For-Review: Create a namespace on ml-serve-eqiad for batch LLM workloads (llm-batch) - https://phabricator.wikimedia.org/T442150#12507907 (10OKarakaya-WMF) cool, thank you @kevinbazira . I'm closing this task as: - llm-batch is created with the target models. -... [11:09:50] 06Machine-Learning-Team (Q2 FY2026-27), 13Patch-For-Review: Create a namespace on ml-serve-eqiad for batch LLM workloads (llm-batch) - https://phabricator.wikimedia.org/T442150#12507912 (10OKarakaya-WMF) 05Open→03Resolved [13:58:37] 06Machine-Learning-Team (Q2 FY2026-27), 06Data-Platform-SRE (2026-10-30 - 2026-11-20), 07Kubernetes, 13Patch-For-Review: Extend KServe's LLMInferenceService by Gateway API Inference Extension and KEDA autoscaling - https://phabricator.wikimedia.org/T436650#12508774 (10DPogorzelski-WMF) Follow-up: services... [15:03:49] 06Machine-Learning-Team (Q2 FY2026-27), 07Essential-Work: Testing scale to 0 in staging. - https://phabricator.wikimedia.org/T443190#12509178 (10isarantopoulos) [15:04:12] 10Lift-Wing, 06Machine-Learning-Team (Q2 FY2026-27), 06Data-Platform-SRE, 06Traffic: Migrate inference and inference-staging to IPIP - https://phabricator.wikimedia.org/T441352#12509179 (10isarantopoulos) [15:04:30] 06Machine-Learning-Team (Q2 FY2026-27), 10CampaignEvents, 06Connection-Team (FY2627 Q2 sprint 3), 05MW-1.47-notes (1.47.0-wmf.24; 2026-10-13), 13Patch-For-Review: Make the LiftWing endpoint configurable, and use the internal one in production - https://phabricator.wikimedia.org/T440360#12509181 (10isarant... [15:04:50] 06Machine-Learning-Team (Q2 FY2026-27), 06DC-Ops, 10ops-eqiad, 06SRE: hw troubleshooting: complete power removal for ml-serve1015 - https://phabricator.wikimedia.org/T441146#12509184 (10isarantopoulos) [15:07:19] 06Machine-Learning-Team (Q2 FY2026-27), 07Essential-Work: Testing scale to 0 in staging. - https://phabricator.wikimedia.org/T443190#12509211 (10OKarakaya-WMF) [16:07:17] 06Machine-Learning-Team (Q2 FY2026-27), 06DC-Ops, 10ops-eqiad, 06SRE, 07Essential-Work: QFY2627 :rack/setup/install ml-serve1016 - https://phabricator.wikimedia.org/T438161#12509695 (10Papaul) We are still working with SM, they requested to provide them with some log files and screenshots, which we did y...