diff --git a/docs/docs/extraction/releasenotes.md b/docs/docs/extraction/releasenotes.md index 9adb164a9..6c696269e 100644 --- a/docs/docs/extraction/releasenotes.md +++ b/docs/docs/extraction/releasenotes.md @@ -2,36 +2,36 @@ This documentation contains the release notes for [NeMo Retriever Library](overview.md). -## 26.03 Release Notes (26.3.0) +## 26.05 Release Notes (26.5.0) -NVIDIA® NeMo Retriever Library version 26.03 adds broader hardware and software support along with many pipeline, evaluation, and deployment enhancements. +NVIDIA® NeMo Retriever Library version 26.05 builds on the 26.03 foundation with a graph-based ingest architecture, expanded multimodal and tabular capabilities, production-oriented service deployment, and documentation aligned to a Helm-first supported path. To upgrade the Helm charts for this release, refer to the [NeMo Retriever Library Helm Charts](https://github.com/NVIDIA/NeMo-Retriever/blob/main/nemo_retriever/helm/README.md). -Highlights for the 26.03 release include: - -- Legacy ingestion repository consolidated under NeMo-Retriever -- NeMo Retriever Extraction pipeline renamed to NeMo Retriever Library -- NeMo Retriever Library now supports two deployment options: - - A new no-container, pip-installable in-process library for development (available on PyPI) - - Existing production-ready Helm chart with NIMs -- Added documentation notes on Air-gapped deployment support -- Added documentation notes on OpenShift support -- Added support for RTX4500 Pro Blackwell SKU -- Added support for llama-nemotron-embed-vl-v2 in text and text+image modes -- New extract methods `pdfium_hybrid` and `ocr` target scanned PDFs to improve text and layout extraction from image-based pages -- VLM-based image caption enhancements: - - Infographics can be captioned - - Reasoning mode is configurable -- Enabled hybrid search with Lancedb -- Added retrieval_bench subfolder with generalizable agentic retrieval pipeline -- The project now uses UV as the primary environment and package manager instead of Conda, resulting in faster installs and simpler dependency handling -- Default TTL for long-running pipeline job state increased from 1–2 hours to 48 hours so long-running jobs (for example, VLM captioning) do not expire before completion -- NeMo Retriever Library currently does not support image captioning via VLM; this feature will be added in the next release -- Documentation: multimodal extraction is covered on one page with an in-page table of contents and redirects from the former per-topic URLs +Highlights for the 26.05 release include: + +- **Graph-based ingest pipeline** — `graph_pipeline` and the graph stage registry are the canonical ingestion path; mode-specific example scripts are consolidated around this model +- **Root CLI** — `retriever ingest` and `retriever query` with NIM URL flags, batch tuning, and LanceDB controls (overwrite/append) +- **Retriever Service v2** — scalable multi-pod architecture with gateway, process isolation, and VectorDB integration +- **Nemotron OCR v2** — default OCR engine with CLI language selectors and unified OCR actors +- **VLM image captioning** — image captioning via vLLM (including Omni caption model profiles); addresses the capability deferred in 26.03 +- **vLLM inference stack** — vLLM-backed text and vision-language embedders, multimodal VL reranker, and torch 2.11 stack for local GPU installs +- **Video retrieval pipeline** — frame extraction, OCR, audio-visual fusion, and text deduplication for video corpora +- **Text-to-SQL** — agent graph and tabular tooling for structured data retrieval +- **Live RAG SDK** — `Retriever.answer()` and optional batch operator graph via LiteLLM (`[llm]` extra) +- **Vector database** — VDB operators integrated directly in the pipeline; custom metadata support; LanceDB hybrid search guidance updated +- **Evaluation** — BEIR-centric evaluation overhaul; `retriever skill-eval` benchmark CLI for the NeMo Retriever skill +- **Packaging** — optional install extras (`[local]`, `[multimedia]`, `[llm]`, `[tabular]`, `[nemotron-parse]`, `[service]`, and others) including slim remote/NIM-only installs on Mac and Windows +- **Audio** — long-audio Parakeet chunking with time-aligned segments; punctuation-based audio segmenting +- **`allow_no_gpu`** — option to skip GPU requirement during ingest for CPU-only experimentation +- **Chunking API** — text splitting moved into `.extract(split_config=...)` +- **Documentation** — Helm-first deployment story; [Docker Compose for local development](https://github.com/NVIDIA/NeMo-Retriever/blob/main/nemo_retriever/docker.md) documented as **unsupported** developer tooling (not a production NIM deployment path) +- **Documentation** — duplicate user-defined stages page removed; UDF and custom stages guidance consolidated in the [graph README](https://github.com/NVIDIA/NeMo-Retriever/tree/main/nemo_retriever/src/nemo_retriever/graph#nemo-retriever-graph) +- **Documentation** — consolidated extraction concepts, ingest workflow, embeddings, and audio/video guides ## Release Notes for Previous Versions +| [26.03](https://docs.nvidia.com/nemo/retriever/26.3.0/extraction/releasenotes/) | [26.1.2](https://docs.nvidia.com/nemo/retriever/26.1.2/extraction/releasenotes/) | [26.1.1](https://docs.nvidia.com/nemo/retriever/26.1.1/extraction/releasenotes/) | [25.9.0](https://docs.nvidia.com/nemo/retriever/25.9.0/extraction/releasenotes/)