OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Reasoning From Scratch

Reasoning From Scratch is a public GitHub repository of Jupyter notebooks and code accompanying the book Build a Reasoning Model (From Scratch). It teaches reasoning-language-model development in PyTorch, starting from a pretrained base model and progressing through methods for improving reasoning.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Reasoning From Scratch — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you want a book-linked, code-first path for understanding reasoning-model development and are prepared to work with notebooks and a Python environment.

Who it is for

  • Readers studying how reasoning language models are developed.
  • Learners who want to work through model-development concepts in PyTorch notebooks.
  • Developers interested in educational examples of inference-time scaling, reinforcement learning, and distillation.

Documented uses

  • Study text generation and evaluation with a pretrained language model.
  • Work through methods for improving reasoning, including self-refinement and inference-time scaling.
  • Explore training approaches including reinforcement learning and GRPO.
  • Study reasoning-model distillation and related evaluation material.
Before you choose

Requirements & limits

The repository is public and licensed under Apache-2.0.

  • The code uses PyTorch and Jupyter notebooks.
  • The book starts with a pretrained Qwen3 base LLM and applies reasoning methods to it.
  • The repository README points to Chapter 2 for Python installation, package management, and environment setup.
  • Hardware and runtime requirements are not specified in the repository overview.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Get the code and locate the chapter notebooks

Download the repository and identify the notebook sequence for studying reasoning-model development.

Before you start

  • A Python coding environment; the repository points to Chapter 2 for setup guidance.

Steps

  1. Download the repository ZIP or clone https://github.com/rasbt/reasoning-from-scratch.

  2. Use the README table of contents to find the chapter notebooks and exercise solutions.

  3. Follow the chapters from pretrained-LLM text generation and evaluation through inference-time scaling, self-refinement, reinforcement learning, and distillation.

Expected outcome

A local copy of the chapter materials and a structured path through the book’s hands-on topics.

Guide limitations

  • The repository overview does not specify hardware or runtime requirements.
  • This is a learning codebase, not a hosted model service.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
7 October 2026
Original website title
rasbt/reasoning-from-scratch · Official homepage title · retained in its source language
How we review sourcesSuggest a correction