Materials Intelligence Recipe knowledge pipeline

Live cumulative operating metrics for the Materials Intelligence Recipe and Recipe Language Model knowledge pipeline. The knowledge base is updated incrementally as new candidate records, qualified papers, PDFs/SI, recipe reports, and RecipeQA pairs enter the system.

Open Exhibition Mode
Candidate Literature
DOIs / Titles / Abstracts
60000+records
open literature intake
Qualified Literature
screened materials papers
20000+papers
scope-qualified corpus
Downloaded PDFs/SI
full text and supplementary information
15000+files
local document archive
Recipe Reports
structured recipe report assets
12000+recipe reports
recipe knowledge assets
Recipe QAs
training and evaluation pairs
120000+pairs
RLM corpus expansion

Literature-to-recipe operating pipeline

These scale bars indicate relative data accumulation across an open-ended operating pipeline. They visualize the current scale of the knowledge system while remaining compatible with continuous growth.

CandidateDOIs, titles, and abstracts collected as the open-ended candidate pool for materials-intelligence literature mining.
60000+ records
QualifiedScope-filtered literature aligned with materials recipes, robotic platforms, digital twins, and AI-driven material discovery.
20000+ papers
PDFs / SIFull-text papers and supplementary information archived for downstream parsing and recipe extraction.
15000+ files
Recipe ReportsStructured recipe reports connecting formulas, processing parameters, performance metrics, and mechanism descriptions.
12000+ recipe reports
RecipeQAQuestion-answer corpus for Recipe Language Model training, evaluation, preference alignment, and mechanistic reasoning.
120000+ pairs

Knowledge assets in Materials Intelligence Recipe

Materials Intelligence Recipe is the structured data layer that turns literature and expert knowledge into reusable assets for search, recommendation, training, and downstream system execution.

Literature scale

Candidate DOIs, titles, abstracts, qualified literature records, and downloaded PDFs/SI form the expandable document base for materials-intelligence knowledge acquisition.

Recipe structuring

Full-text and supplementary materials are converted into structured recipe reports containing formulas, process parameters, performance metrics, and mechanism-aware annotations.

Recipe-QA assets

Structured recipe reports are transformed into RecipeQA pairs that support Recipe Language Model training, evaluation, preference alignment, and scientific interaction.

Materials Intelligence Recipe
Expert-mechanism entry, AI recipe recommendation, and the interaction layer for materials-intelligence workflows.
Open in New Window