@metawake

Detecting Policy Puppetry in LLM Prompts: A Simple, Transparent Approach

Apr 27, 2025 · 3 min read · There’s a lot of excitement these days about large language models (LLMs) and the new opportunities—and risks—they bring. As someone who’s spent a fair bit of time in the NLP and LLM trenches, I’ve seen firsthand how creative people can get when tryi...

Flexible Entity Extraction for NLP Pipelines: Introducing nergrep

Apr 26, 2025 · 4 min read · As an AI/NLP/ML platform engineer, I often need to prototype and iterate on text analytics pipelines. Entity extraction is a frequent, foundational step—whether I'm building quick R&D experiments or shaping components for production. I wanted a tool ...

Best Practices for Prompt Modifier Shaping: Navigating Model Behavior and Responsiveness

Apr 21, 2025 · 3 min read · Some models resist the formatting, length, or tone modifiers we give them. But what if that's not a bug, but a feature we can work around? Context and Purpose In recent experiments, we explored how prompt modifiers influence the length, style, and st...

Prompt Shaping: Measuring the Impact of Prompt Modifiers on Output Size and Format

Apr 21, 2025 · 4 min read · What if you could guide large language models (LLMs) to output shorter, clearer, or more structured responses just by adjusting the prompt? This experiment set out to answer a simple yet valuable question: Can we shrink or reshape LLM output just by...

Semantic Prompt Compression: Reducing LLM Costs While Preserving Meaning

Apr 14, 2025 · 3 min read · View the open-source project on GitHub The Challenge: Every Token Costs In the world of Large Language Models (LLMs), every token comes with a price tag. For organizations running thousands of prompts daily, these costs add up quickly. But what if w...