Tag
This paper compares fine-tuned BERT (gbert-large) with few-shot LLM prompting (Llama 4 Maverick) for detecting threat and solution framing in German climate news sentences. BERT achieves higher F1 scores (0.83 vs 0.78), and an ablation study shows that providing preceding sentence context improves performance.
A prompt engineering technique that asks ChatGPT to first identify assumptions, missing information, and common mistakes before answering, to get more thoughtful and tailored responses.
This paper studies performance prediction for symbolic (e.g., Python) and prompt programs using a Bayesian coin-flip model, finding that symbolic programs have all-or-nothing performance while prompt programs have diffuse priors, and introduces RAP (Retrieved Approximate Prior) for performance prediction.
The article argues that treating AI as an equal partner yields better results for complex tasks, while precise prompting is still suitable for technical tasks.
The author details their decision to exclude LLMs from generating final fact-check verdicts in favor of a hybrid architecture that uses LLMs for data extraction and a deterministic Python layer for scoring, citing issues with stochastic instability and auditability.
Andrej Karpathy suggests prompting LLMs to structure responses as HTML for better visualization and predicts AI output will evolve from text to interactive neural videos.
DiZiNER is a framework that uses disagreement between multiple LLMs to refine task instructions for zero-shot named entity recognition, achieving state-of-the-art results on 14 out of 18 benchmarks and significantly reducing the performance gap between zero-shot and supervised systems.