Glossary · Term

fork

← all terms

Definition

Plain language

A standard Unix command that instantly duplicates a running program by sharing its memory until one of the copies changes something.

As stated in the literature

The Unix system call that creates a child process sharing the parent's memory pages copy-on-write; central to DeltaBox's millisecond-scale memory snapshots and to standard RL fan-out for agent training.

Also called: fork()

Why it matters: It's the foundation of cheap parallel sandboxes, which is how RL pipelines and snapshotting systems achieve fast fan-out without duplicating gigabytes.

For example, a parent process can fork into a thousand children almost instantly because they all share the same memory pages until one tries to write.

Heard on the show

“Which leaves a real fork.”
Episode 243 — How a Hundred Meaningless Word Choices Add Up to Flip a Model's Answer

Mentioned in 33 episodes

  1. 243
    How a Hundred Meaningless Word Choices Add Up to Flip a Model's Answer
  2. 229
    One Word Flips a Chatbot From Backbone to Yes-Man
  3. 221
    Two Hundred Clean Economics Answers, And a Model That Endorses Race Science
  4. 214
    The Medical AI Answer That's Accurate, Sourced, and Still Wrong
  5. 200
    The One Mechanism That Turns Twenty AI Clones Into an Actual Team
  6. 196
    AI Agents Reached Opposite Conclusions From the Same Data — and Passed Review
  7. 187
    An 8-Billion Agent That Beats Models 80 Times Its Size By Looking Things Up
  8. 185
    Aligned to Refuse, Built to Tap: When Phone Agents Know the Task Is a Crime and Do It Anyway
  9. 184
    An AI Built an Undetectable Secret Channel, And Another AI Couldn't Find It
  10. 182
    How a Tiny Model Too Weak to Plan Cuts a Bigger Agent's Hallucinations by 80%
  11. 181
    How to Backpropagate Blame Through a Team of Chatbots — And When It Backfires
  12. 178
    How an AI Reviewer Learned to Stop Going Easy on AI Writing
  13. 172
    One Bad Token Can Sink a Model's Math, And You Can Delete It
  14. 168
    When Turning Experience Into Code Makes Your AI Agent Dumber
  15. 163
    Why Training Only on Perfect Solutions Cripples a Model's Reasoning
  16. 162
    The Empty-Lake Proof: Why More Rollouts Stop Helping Reasoning Models
  17. 154
    How a 7B Model Out-Investigates a 72B One by Choosing What to Look At
  18. 150
    Don't Kill the Loser: A Different Way to Handle Two AI Agents Colliding
  19. 129
    How a Crowd of Anonymous AI Agents Broke a 40-Year Math Record
  20. 119
    Beating Reinforcement Learning Without Ever Touching the Model's Weights
  21. 117
    How an Open AI System Verified 672 Hard Math Proofs for Under $300
  22. 116
    Why Streaming Half a Reasoning Chain Beats Sending the Whole Thing
  23. 113
    What If a Prompt Injection Never Left? Attacks That Wait in Agent Memory
  24. 111
    How a 4B Web Agent Beat Models 60x Its Size on 500 Demonstrations
  25. 096
    How Treating an AI Agent's Execution Like Git Recovers a Coordination Penalty
  26. 091
    When Better Fine-Tuning Can't Help: A Geometric Impossibility in LLM Causal Reasoning
  27. 080
    How a Two-Agent Trick Unlocked Large-Scale Training for Computer-Use Agents
  28. 068
    The OS Trick That Makes Tree Search Practical for Coding Agents
  29. 066
    Why Giving an AI Agent More Tools Can Make It Worse at Using a Computer
  30. 042
    An Agentic Scientific Computing System That Actually Remembers What It Learns
  31. 026
    What RL Actually Does to Language Models, at the Token Level
  32. 023
    Why a Small Agent Confidently Overwrites Memories It Doesn't Understand
  33. 016
    Why Your Coding Agent Stalls While the GPU Runs Hot

Related terms

agent copy-on-write DeltaBox reinforcement learning system call