A LITTLE CONTEXT FOR THE COLLECTION
Open doors.
Traceable sources.
The Learning Garden combines an editorial collection with a larger index of open learning materials. It is a personal project by Shivji Agnihotri, created to make machine learning easier to explore and share.
Two ways to explore
The visual concepts, bookshelf, and recommended paths are selected for clarity and learning value. The full library goes wider: individual tutorials, lessons, reading material, and research records gathered from official sites and maintained open collections. An indexed entry is a place to explore, not a claim that every learner needs it.
Where the links come from
Books link to their authors, authorized publishers, universities, or official repositories. Tutorials and course lessons are collected from original providers. Research records use published open metadata and link to the original paper. Each card includes its provider and, where useful, the source used to discover it.
- Book-by-book source and access notes
- Selected courses, guides, and official access checks
- Current library counts and import date
- Catalogue data and collection notes on GitHub
- The scripts used to refresh the collection
What “free” means here
The linked reading, videos, documentation, or research papers can be accessed without purchasing the material. Some providers also offer paid certificates, optional cloud compute, or commercial products. Those services are separate from the free learning links. A small number of older author sites use HTTP; they are retained as primary sources rather than replaced with unofficial copies.
Checks, dates, and limitations
The curated links were reviewed in October 2026. The larger index records its import date and provenance in the collection notes. Verification varies by source: a resource may be directly checked by HTTP, discovered in an official sitemap, or confirmed through an open research metadata record. Links can move, and a successful page response does not assess teaching quality.
Research papers may be preprints and may not have been peer reviewed. “Advanced” reflects the mathematical reading expected of research, while “Not level-specific” means the source does not assign a level. Topic labels are browsing aids and may not cover every subject within a resource.
10,000 additional AI resources
The expanded library includes books, practical scikit-learn tutorials, and individual research articles from ACL Anthology. Research makes up most of the additions and is marked Advanced. Beginner and Intermediate identify teaching material and practical prerequisites; these labels are editorial estimates. Use the subject filter to narrow topics such as retrieval, reasoning, evaluation, or neural networks. Expansion sources and methodology and the import audit explain what was fetched and what was source-indexed.
50,000 resources and learning to build with LLMs
The library now includes English research from international proceedings and practical LLM guides, books, articles, and videos. Explore loading an open model, training, fine-tuning, running locally, quantization, and deployment using the skill shortcuts. Research makes up most of the collection; use the format and level filters for teaching material. Worldwide and LLM sources, counts, and checks document the expansion.
Research language checks use English publisher pages and abstracts; individual PDFs and video audio have not been reviewed one by one. Geographic coverage records conference venue locations across six continents, rather than author nationality. Online or unspecified locations remain explicitly unspecified.
Video previews
All video entries use preview thumbnails supplied by their original providers, saved locally for reliable display. A selected course without source artwork has a garden illustration. Covers open the lessons on their original host. The creators retain ownership of their preview images; image sources and dates are recorded with the collection.
Made to be shared
Use “Copy this collection link” to share the library with its search and filters. Each concept also has its own shareable address. Hearts save items in your browser on this device; they are not uploaded or included in shared links. The interactive examples run locally and use small, explained datasets.
Help the garden grow
Found a broken link, a misleading label, or a useful open resource? Suggest a correction on GitHub. Original books, courses, and papers remain with their authors. This site indexes links and metadata; it does not rehost their contents.