🚀
Codex Editor
  • Project Overview
    • Welcome to Codex
    • Codex Editor
      • Features
      • What was Project Accelerate?
        • Project Philosophy
        • Steering Committee
    • Vision
      • Powerful Tools and Simplicity
      • Streamlining the Translation Process
      • Unopinionated Assistance
    • Architecture
    • Frequently Asked Questions
    • Translator Requirements
    • Roadmap
    • Why VS Code?
    • Multimodality
  • Translator's Copilot
    • What is Translator's Copilot?
    • Information Management
      • Resource Indexing
      • Greek/Hebrew Insights
      • Chat with Resources
      • Project-Based Insights
    • Translation Assistance
      • Translation Drafting
        • Prioritizing Semantics over Structures
        • Reranked Translation Suggestions
      • Quality Checking
        • Token matching evaluation
        • Few-shot evaluation
        • Fine-tuned Bible QA model
        • Character and word n-gram evaluation
        • Simulated GAN evaluation
        • Linguistic Anomaly Detection (LAD)
      • Back Translation
    • Orchestration
      • Translation Memory
      • Multi-Agent Simulations
      • Drafting linguistic resources
    • Intelligent Functions Library
  • Development
    • Codex Basics
      • Projects
      • The Editor
      • Extensions
        • Rendered Views
        • Language Servers
        • Custom Notebooks
        • Global State
    • Experimental Repositories
Powered by GitBook
On this page
  1. Translator's Copilot
  2. Translation Assistance

Back Translation

PreviousLinguistic Anomaly Detection (LAD)NextOrchestration

Last updated 1 year ago

A lot of thought needs to go into ways that back-translation can be constrained from hallucinating.

Mechanisms for constraining back-translation and avoiding hallucination will range from rigid to flexible. On the rigid end of the spectrum will be methods like Token matching evaluation and valid_token constraints. On the more flexible end, we could try pulling in the most similar verses to the back-translation draft and comparing the target-language 'source' in each case. If the model begins to hallucinate, it will probably not match well with existing translations of similar verses.

Cipher-based completions (see discussion ) might be a way we can mask the tokens from the LLM, but we would need to be sure to offer enough in-context data to ensure useful patterns are detected.

here