All pull requests

Pull requests

• feat(batch): opt-in elastic actor pools from heuristic bounds

  #2775·jioffe502 opened Oct 7, 2026Collaborator·3 tasks done·

  ·
  8

  

• Adds re-indexing NB

  #2773·mahikaw opened Oct 6, 2026Collaborator·3 tasks·

  ·
  2

• feat(ingest): stage batch ingest rows to Parquet from Ray workers (--stage-dir)

  #2771·jioffe502 opened Oct 5, 2026Collaborator·

  ·

• feat: add scalable Parquet vector indexing with LanceDB and Elasticsearch/cuVS

  #2766·jperez999 opened Oct 5, 2026Collaborator·

  ·
  10

• feat(vdb): load cached Arrow vectors through the LanceDB writer

  #2765·jioffe502 opened Oct 5, 2026Collaborator·

  ·
  6

• feat(extraction): support Nemotron Parse handoffs to Curator

  #2764·KyleZheng1284 opened Oct 5, 2026Collaborator·3 tasks done·

  ·
  13

• Add ViDoRe v3 HR retrieval evaluation notebook

  #2763·ChrisJar opened Oct 2, 2026Collaborator·2 of 3 tasks·

  ·
  2

• Add retryable retriever unload chain with explicit vLLM shutdown

  #2762·edknv opened Sep 30, 2026Collaborator·3 tasks·

  ·
  7

• Keep doubled emphasis markers out of empty tags

  #2761·SashaMIT opened Sep 30, 2026·

  ·
  3

• Expose agent LLM timeout through typed retrieval configuration

  #2760·edknv opened Sep 30, 2026Collaborator·3 tasks·

  ·
  9

• Use descriptive names for stored images

  #2759·JiKaiHao opened Sep 30, 2026·3 tasks done·

  ·
  1

  5

• Add OfficeQA Pro v2 BEIR benchmark

  #2758·edknv opened Sep 29, 2026Collaborator·3 tasks·

  ·
  10

• docs(batch): add measured four-L40S allocation example

  #2757·jioffe502 opened Sep 29, 2026Collaborator·

  ·

• feat(batch): support native bounded OCR actor pools

  #2756·jioffe502 opened Sep 29, 2026Collaborator·

  ·
  19

• docs: clarify default extraction method behavior

  #2755·efegokdemir opened Sep 29, 2026·

  ·
  1

  2

• fix: render PDF pages before Nemotron Parse in the in-process pipeline

  #2751·v0ropaev opened Sep 28, 2026·3 tasks done·

  ·
  13

• fix(sdk): SDK/CLI discrepancy fixes

  #2750·jioffe502 opened Sep 28, 2026Collaborator·3 tasks done·

  ·
  12

• feat(profiling): add semantic batch NVTX ranges

  #2748·jioffe502 opened Sep 23, 2026Collaborator·3 tasks done·

  ·
  21

• bench: add experimental cuDF direct transport adapter

  #2742·jperez999 opened Sep 22, 2026Collaborator·3 tasks done·

  ·
  9

• feat: support bounded elastic embedding actor pools in batch mode

  #2741·jioffe502 opened Sep 22, 2026Collaborator·

  ·
  8

• Compact built-in embedding rows for summary uploads

  #2737·jioffe502 opened Sep 22, 2026Collaborator·2/2·2 of 3 tasks·

  ·
  5

• fix(helm): allow NVFP4 artifact selection for SM120 embedding caches

  #2731·KyleZheng1284 opened Sep 21, 2026Collaborator·3 tasks done·

  ·
  2

• fix(extraction): validate OCR dependencies while preserving audio opt…

  #2727·KyleZheng1284 opened Sep 18, 2026Collaborator·3 tasks done·

  ·
  3

• feat: Add Qdrant as an alternate vector search backend

  #2708·anush008 opened Sep 17, 2026·3 tasks done·

  ·
  8

• fix(ci): preserve Docker image tag for publish

  #2702·jperez999 opened Sep 16, 2026Collaborator·3 tasks·

  ·
  2