All pull requests

Pull requests

• [tune] Respect max_concurrent_trials for restored trials

  #67670·vineethsaivs opened Nov 1, 2026Contributor·

  ·
  2

• [data] Prune Parquet row groups with bloom filters for equality predicates

  community-contributiondata
  #67669·xuzifu666 opened Nov 1, 2026·

  ·
  5

• [Data] Preserve nulls and handle near-zero IQR in RobustScaler

  community-contributiondata
  #67668·deep1blue opened Nov 1, 2026·5 of 7 tasks·

  ·
  6

• [RLlib] Apply action transforms consistently in TorchSquashedGaussian

  community-contributionrllib
  #67667·Whning0513 opened Nov 1, 2026Contributor·

  ·
  3

• [Autoscaler][v2] Ignore max_concurrent_launches for KubeRay

  corekubernetes
  #67666·machichima opened Nov 1, 2026Contributor·

  ·
  1

  8

• [data][llm] Keep the prompt in vLLM error rows

  community-contributiondata
  #67665·arun-elr opened Nov 1, 2026·

  ·
  4

• [Data] Vectorize ValueCounter column combination via VectorizedAggregateFnV2

  community-contributiondataperformance
  #67664·deep1blue opened Nov 1, 2026·5 tasks done·

  ·
  5

• [RLlib] Exclude masked timesteps from GAE normalization

  community-contributionrllib
  #67663·Whning0513 opened Oct 31, 2026Contributor·

  ·
  1

  1

• [Data] Persist BlockMetadata.num_rows and batch ray.get in _calculate_blocks_rows

  community-contributiondata
  #67662·deep1blue opened Oct 31, 2026·1 task done·

  ·
  3

• [Data] Add fill_null expression method

  community-contributiondata
  #67661·surawut-jirasaktavee opened Oct 31, 2026·

  ·
  5

• [Data] Fix flaky test_torch_tensor_utils: raise Bazel size to medium

  #67660·AarryaSaraf opened Oct 31, 2026Contributor·

  ·

• [Data] Fix flaky test_data_iterator_checkpointer: wait for a readable checkpoint file

  #67659·AarryaSaraf opened Oct 31, 2026Contributor·

  ·

• [core][util] Fix ray.util.queue.Queue put/get with timeout=0

  community-contributioncore
  #67658·deep1blue opened Oct 31, 2026·4 of 8 tasks·

  ·
  3

• [Data] Support Std and AbsMax in GPU hash aggregation

  community-contributiondata
  #67657·WangGuangxin opened Oct 31, 2026·

  ·
  6

• [Data] Push AND conjuncts on passthrough columns below a Project

  community-contributiondata
  #67656·sahil-mangla opened Oct 31, 2026·

  ·
  1

  4

• [data][llm] Split SGLang engine stage and processor into per-task classes

  community-contributiondata
  #67655·fortunecookiee opened Oct 31, 2026Contributor·

  ·
  4

• [Tune] Preserve errored trials when importing evaluated results

  community-contributiontune
  #67654·MakerFlyFly opened Oct 31, 2026·

  ·
  1

  1

• [Data][CI] Run the Ray Data pyrefly check in pre-commit

  datadevprod
  #67652·goutamvenkat-anyscale opened Oct 31, 2026Contributor·

  ·
  1

• [Train] Shut down failed worker group before sizing replacement

  community-contributiontrain
  #67651·ryanaoleary opened Oct 31, 2026Contributor·

  ·
  5

• [Data] Estimate PyArrow's peak memory decoding a Parquet row group from its footer [5/5]

  #67650·AarryaSaraf opened Oct 31, 2026Contributor·

  ·
  1

• [Data] Log the conflicting column and types on schema unification error

  datago
  #67647·ayushk7102 opened Oct 31, 2026Contributor·

  ·
  8

• [Data] Drain Iterator to Fix Flakiness in Rate-Based Cluster Autoscaler Test

  datago
  #67646·rayhhome opened Oct 31, 2026Contributor·

  ·
  3

  

• [Data] Skip the S3 HEAD request on each Parquet file open by passing listed file sizes

  datago
  #67645·AarryaSaraf opened Oct 31, 2026Contributor·

  ·
  5

• Bump follow-redirects in /python/ray/dashboard/client

  coredependenciesjavascript
  #67644·dependabot[bot] opened Oct 30, 2026Bot·

  ·
  1

• [Data] Size Parquet read tasks by their decoded Arrow bytes in DSv2 [4/5]

  data
  #67642·AarryaSaraf opened Oct 30, 2026Contributor·

  ·
  1