How deeply does Build a Reasoning Model (From Scratch) cover reinforcement learning with human feedback?

Explains candidate generation, preference collection, reward-model training, scoring, and RL updates, but defers algorithmic implementation details.

Reinforcement learning with human feedback

As explained in Build a Reasoning Model (From Scratch)

A training approach that uses human preference labels to optimize a language model toward responses humans prefer.

Reinforcement learning with human feedback (RLHF) is a preference-tuning method that uses human rankings of model outputs to optimize a language model toward responses people prefer. Unlike pretraining and supervised fine-tuning, it optimizes complete responses rather than only next-token predictions. section 6.1.1 section 1.2

Implement RLHF as a two-stage workflow:

Generate multiple candidate responses for each prompt, then collect human rankings or pairwise preferences. section 6.1.1
Train a reward model, itself typically an LLM, to assign preference scores that reflect the collected rankings. section 6.1.1
Generate target-model outputs, score them with the reward model, and update the target LLM with an RL objective that encourages preferred behavior. section 6.1.1

Choosing a laptop

For a programming-and-everyday-use prompt, one response lists many possible laptop features, while another prioritizes a fast CPU, at least 16 GB of RAM, an SSD, keyboard quality, and battery life. Human annotators may rank the second response higher because it is more concrete and tailored to the stated use case. section 6.1.1

RLHF requires an additional learned reward model, which can be expensive to train. RL with verifiable rewards (RLVR) replaces that model and human annotation with deterministic checks, such as mathematical answer verification, and therefore uses a simpler training loop. section 6.1.2

How Build a Reasoning Model (From Scratch) covers it

Cover of Build a Reasoning Model (From Scratch)

liveBook manning.com

Read Build a Reasoning Model (From Scratch) on liveBook Get the book at manning.com

Also explained in

liveBook manning.com

Reinforcement Learning from Human FeedbackReinforcement learning from human feedback (RLHF) is a model-alignment technique that optimizes behavior using human preference signals, often through a learned reward model or direct preference objective.
liveBook manning.com

LLMs in ProductionReinforcement learning from human feedback is a model-training approach that optimizes behavior using preference or evaluation signals supplied by people.
liveBook manning.com

How Large Language Models WorkReinforcement learning from human feedback is a technique that optimizes a model's behavior using human evaluations as a source of reward or preference information.
liveBook manning.com

Generative AI in ActionReinforcement learning from human feedback is a model-training technique that uses human preferences as feedback to optimize generated behavior.
liveBook manning.com

The AI Pocket BookA training method that uses human judgments as feedback to guide reinforcement learning and improve model outputs.
liveBook manning.com

Grokking AI ApplicationsReinforcement learning from human feedback is a training technique that optimizes model behavior using human evaluations of its responses.
liveBook manning.com

Natural Language Processing in Action, Second EditionReinforcement learning with human feedback is a model-training technique that uses human evaluations to optimize a system's behavior.
liveBook manning.com

Introduction to Generative AI, Second EditionReinforcement learning from human feedback (RLHF) is a technique that uses human evaluations or preferences to estimate rewards and optimize model behavior.
liveBook manning.com

Deep Learning with R, Third EditionReinforcement learning with human feedback is a training approach that optimizes a model using human preference judgments converted into reward signals.
liveBook manning.com

Deep Learning with Python, Third EditionA training setup that uses human preference feedback to build a reward signal for improving a model’s outputs.
liveBook manning.com

Transformers in ActionReinforcement learning from human feedback is a model-optimization technique that uses human preference feedback to define or improve a reward signal.
liveBook manning.com

First-Party Data ActivationReinforcement learning from human feedback is a model-training technique that uses human evaluations to shape a system’s behavior.
liveBook manning.com

Software Testing with Generative AIA model-alignment technique that uses human feedback to guide or refine a model's behavior.

Compare how 14 books explain reinforcement learning with human feedback →

Related concepts

reward modeluses · A reward model is a learned model that assigns scores to outputs, actions, or outcomes to represent their desirability according to preferences or evaluation criteria.
Proximal Policy Optimizationuses · Proximal Policy Optimization is a policy-gradient reinforcement-learning algorithm that limits policy updates to keep successive policies close and promote stable learning.
reinforcement learninguses · Reinforcement learning is a machine-learning paradigm in which an agent learns actions or a policy through interaction with an environment and evaluative feedback, such as rewards.
Direct Preference Optimizationcontrasts with · Direct Preference Optimization is a technique for training a model directly from preference data, such as preferred and dispreferred outputs, without separately fitting a reward model.
RLAIFalternative to · Reinforcement learning from AI feedback is a training or alignment technique that uses AI-generated evaluations or preferences as feedback to optimize a model or agent.
ChatGPTused by · ChatGPT is OpenAI’s conversational AI service, which lets users interact with generative AI models and related tools through natural-language prompts.

This explainer was generated by Manning's knowledge graph from the text of Build a Reasoning Model (From Scratch). The section links point to the parts of the book it draws on.