Open
Conversation
kheiss-uwzoo
requested review from
jdye64,
jperez999 and
sosahi
and removed request for
a team
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Overview
QA review branch that brings in upstream main and applies bug-queue doc and tooling updates. The diff is ~2,300 insertions and ~1,200 deletions across 66 files, touching docs, NeMo Retriever, Helm, the test harness, and retrieval-bench.
Documentation
Extraction docs: Updates across audio, benchmarking, quickstart (guide + library mode), support matrix, content/custom metadata, FAQ, CLI reference, Python API reference, VLM embed, user-defined functions/stages, and v2 API guide.
Naming/links: Consistent use of “NeMo Retriever Library” (replacing “NVIDIA Ingest” / “nv-ingest”) and fixes for support matrix and related links (e.g. RIVA).
Helm: README and values updates; table additions for nimOperator.rerankqa and nimOperator.ocr; nemotron rebranding.
NeMo Retriever
Markdown API: to_markdown() returns None for empty results; markdown I/O and tests adjusted (including test_io_markdown, test_html_convert, test_txt_split).
Image support: Docs for ingesting image files (batch and in-process), including extract_image_files and --input-type image.
Text chunking: .split() for token-count–based chunking (#1547).
Audio: Batched audio extraction improvements; Parakeet CTC ASR and ASR actor updates.
Build/install: Retriever installed as part of Docker build; get_hf_revision removed from code outside nemo_retriever/ (#1612).
Release: Version handling and PyPI wheel naming; NeMo Retriever LICENSE added.
Helm & Harness
Helm: RTX PRO 4500 override and obj-det warmup batch size override; reranker and OCR NIM table docs.
Harness: Wait for healthy reranker when needed for recall; retry for managed Helm port-forwards; docker-compose and Helm service manager improvements; JP20 recall config cleanup and readiness logging.
Retrieval-bench
Pipeline and modality handling improvements; refactors for retriever singletons (ColeEmbed, HF dense, Nemotron ColeEmbed VL v2, Nemotron Embed VL dense).
New BRIGHT agentic submission (bright_agentic.md, sdg.png).
CI / Release
Perform-release: Workflow and release-helm updates; reusable PyPI build/publish and release-helm workflow changes.
Misc: Redis TTL default increased to 48h for VLM captioning; in-process extract fixes for txt and reranker; source_id in LanceDB schema; rerank and release-related fixes.