[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:lift":3},"\u003Cp align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fdatalab-to\u002Flift\u002FHEAD\u002Fassets\u002Fdatalab-logo.png\" alt=\"Datalab Logo\" width=\"150\" \u002F>\n\u003C\u002Fp>\n\u003Ch1>Datalab\u003C\u002Fh1>\n\u003Cp align=\"center\">\n  \u003Cstrong>State of the Art models for Document Intelligence\u003C\u002Fstrong>\n\u003C\u002Fp>\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fopensource.org\u002Flicenses\u002FApache-2.0\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FCode%20License-Apache_2.0-green.svg\" alt=\"Code License\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.datalab.to\u002Fpricing\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FModel%20License-OpenRAIL--M-blue.svg\" alt=\"Model License\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FKuZwXNGnfH\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20us-5865F2?logo=discord&logoColor=white\" alt=\"Discord\" \u002F>\u003C\u002Fa>\n\u003C\u002Fp>\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.datalab.to\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FHomepage-datalab.to-blue\" alt=\"Homepage\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdocumentation.datalab.to\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocs-Read%20the%20docs-blue\" alt=\"Docs\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.datalab.to\u002Fplayground\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPlayground-Try%20it-orange\" alt=\"Public Playground\" \u002F>\u003C\u002Fa>\n\u003C\u002Fp>\u003Chr \u002F>\u003Ch1>lift\u003C\u002Fh1>\n\u003Cp>lift extracts structured JSON from PDFs and images by passing a schema. It's a 9B vision model that returns a JSON object matching your schema, with schema-constrained decoding guaranteeing valid output.\u003C\u002Fp>\n\u003Cp align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fdatalab-to\u002Flift\u002FHEAD\u002Fassets\u002Fextraction-example.png\" alt=\"lift extracting schema-aligned JSON from an invoice\" width=\"48%\" \u002F>\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fdatalab-to\u002Flift\u002FHEAD\u002Fassets\u002Faccuracy.png\" alt=\"Extraction accuracy benchmark\" width=\"48%\" \u002F>\n\u003C\u002Fp>\u003Ch2>Try lift on Datalab\u003C\u002Fh2>\n\u003Cp>Our managed platform runs improved extraction with higher accuracy than the open weights, plus per-field verification, citations, and confidence scores.\u003C\u002Fp>\n\u003Cp>If you have high volume workloads, we offer a batch processing service that has processed 1B+ pages per week.\u003C\u002Fp>\n\u003Cp>Get started with \u003Cstrong>$20 in free credits per month\u003C\u002Fstrong> — \u003Ca href=\"https:\u002F\u002Fwww.datalab.to\u002F?utm_source=gh-lift\" rel=\"nofollow ugc noopener\">sign up\u003C\u002Fa> - takes under 30 seconds - or try lift in our \u003Ca href=\"https:\u002F\u002Fwww.datalab.to\u002Fplayground?utm_source=gh-lift\" rel=\"nofollow ugc noopener\">public playground\u003C\u002Fa>.\u003C\u002Fp>\n\u003Cp>Commercial self-hosting requires a license — see \u003Ca href=\"#commercial-usage\" rel=\"nofollow ugc noopener\">Commercial usage\u003C\u002Fa>. For on-prem licensing, \u003Ca href=\"https:\u002F\u002Fwww.datalab.to\u002Fcontact?utm_source=gh-lift-onprem\" rel=\"nofollow ugc noopener\">contact us\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2>Features\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>Extract structured data from documents\u003C\u002Fli>\n\u003Cli>Pass any JSON schema\u003C\u002Fli>\n\u003Cli>Handles multi-page documents in a single pass, including values that span pages\u003C\u002Fli>\n\u003Cli>Two inference modes: local (HuggingFace) and remote (vLLM server)\u003C\u002Fli>\n\u003Cli>CLI for single files, inline schemas, or whole directories\u003C\u002Fli>\n\u003Cli>Schema Studio: a Streamlit app to build, save, and test schemas against your documents\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Quickstart\u003C\u002Fh2>\n\u003Cp>The easiest way to start is with the CLI tools:\u003C\u002Fp>\n\u003Cpre>\u003Ccode class=\"language-shell\">pip install lift-pdf\n\n# With vLLM (recommended, lightweight install)\nlift_vllm\nlift_extract input.pdf .\u002Foutput --schema schema.json\n\n# With HuggingFace (requires torch)\npip install lift-pdf[hf]\nlift_extract input.pdf .\u002Foutput --schema schema.json --method hf\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Ch2>Benchmarks\u003C\u002Fh2>\n\u003Cp>Evaluated on a 225-document extraction benchmark (6–64 pages per document, ~11,000 scored fields) with adversarial cases planted throughout: cross-page values, exhaustive lists, fields that must be left null, near-miss distractors, multi-source aggregation. Scoring is deterministic exact-match against ground truth (numeric tolerance, normalized strings).\u003C\u002Fp>\n\u003Cp>All models receive the same rendered page images, and extract each document in a single pass.\u003C\u002Fp>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Model\u003C\u002Fth>\n\u003Cth>Size\u003C\u002Fth>\n\u003Cth>Field accuracy\u003C\u002Fth>\n\u003Cth>Full-document accuracy\u003C\u002Fth>\n\u003Cth>Median latency*\u003C\u002Fth>\n\u003Cth>Features\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>Datalab API\u003C\u002Ftd>\n\u003Ctd>—\u003C\u002Ftd>\n\u003Ctd>95.9%\u003C\u002Ftd>\n\u003Ctd>44.4%\u003C\u002Ftd>\n\u003Ctd>30.8s\u003C\u002Ftd>\n\u003Ctd>Citations + Verification\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Gemini Flash 3.5\u003C\u002Ftd>\n\u003Ctd>—\u003C\u002Ftd>\n\u003Ctd>91.3%\u003C\u002Ftd>\n\u003Ctd>40.0%\u003C\u002Ftd>\n\u003Ctd>28.1s\u003C\u002Ftd>\n\u003Ctd>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>\u003Cstrong>lift\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>9B\u003C\u002Ftd>\n\u003Ctd>\u003Cstrong>90.2%\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003Ctd>20.9%\u003C\u002Ftd>\n\u003Ctd>9.5s\u003C\u002Ftd>\n\u003Ctd>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Azure Content Understanding\u003C\u002Ftd>\n\u003Ctd>\u003C\u002Ftd>\n\u003Ctd>\u003C\u002Ftd>\n\u003Ctd>\u003C\u002Ftd>\n\u003Ctd>\u003C\u002Ftd>\n\u003Ctd>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\n",1791762981890]