XML
PulseAugur coverage of XML — every cluster mentioning XML across labs, papers, and developer communities, ranked by signal.
6 day(s) with sentiment data
-
LLM Prompting: Markdown vs. XML Tags for Structure
Large Language Models (LLMs) can be prompted using either Markdown or XML-style tags to delineate different sections of input. Markdown is generally preferred for its readability and ease of maintenance, especially for …
-
LLMs to assist farm machinery via offline operator manuals
Researchers have developed a method for enabling large language models (LLMs) to provide offline, multilingual assistance for agricultural machinery by synchronizing operator manuals over ISO 11783-based CAN buses. The …
-
11 Production-Ready Prompt Engineering Practices Detailed
This article outlines eleven essential practices for effective prompt engineering in production environments, emphasizing the need for robust, repeatable methods over one-off tricks. Key recommendations include being sp…
-
AI document translation's future lies beyond the model, says expert
The author argues that in 2027, the specific AI model used for document translation will be less important than the surrounding pipeline. Current tools often fail because they treat documents as simple strings, neglecti…
-
AI harness engineering significantly narrows model gap in legislative XML conversion
A new research paper explores the effectiveness of "harness engineering" in improving the accuracy of AI models for converting legislative documents into XML. The study found that the deterministic tooling built around …
-
Claude AI's System Prompts Explained: Using XML Tags for Better Control
This article explains the use of XML tags within Claude's system prompts. It details how these tags help structure instructions for the AI, enabling more precise control over its responses and behavior. The piece also p…
-
LLM structured output performance depends on schema design, not structure itself
A new arXiv paper challenges the notion that structured output formats like JSON or XML inherently degrade large language model (LLM) performance. Researchers found that the accuracy loss, previously termed the 'structu…
-
AI agents audit blog posts; PPTX converter misses SmartArt text
A user tested two AI subagents on their AI-assisted blog posts, finding that one agent produced a score of 15.1 out of 25. The user also encountered an issue where a PPTX converter failed to process SmartArt text, which…
-
Pretty JSON uses 3x more tokens than CSV in LLM prompts, analysis finds
A recent analysis compared the token usage of seven data formats when used in LLM prompts, finding that pretty-printed JSON consumes approximately three times more tokens than CSV. This difference is attributed to repea…
-
O-Funnel library offers lossless data extraction from evolving documents
Researchers have developed O-Funnel, a new Python library designed to extract data from heterogeneous and evolving documents without manual pattern writing. O-Funnel separates structural capture from value extraction, t…
-
Developer creates tool to render Android UI layouts with mock data
A developer has created a tool called `android-ui-renderer-mcp` that can render Android XML layouts into PNG images and provide view tree data. This tool goes beyond simple layout rendering by incorporating elements lik…
-
New benchmark suite targets formal verification of industrial PLC programs
Researchers have developed a new benchmark suite for formally verifying IEC 61131-3 ladder diagram programs, addressing a lack of standard evaluation tools in the field. This suite includes 50 programs across ten indust…
-
RAG chunking strategies impact LLM retrieval performance
Optimizing retrieval-augmented generation (RAG) systems requires careful consideration of chunking strategies, as embedding quality directly impacts performance. For text, a chunk size between 256-512 tokens, with 10-20…
-
Guide to Structuring Claude Opus Prompts with XML Tags
This article provides a guide on how to effectively structure prompts for Anthropic's Claude Opus model using XML tags. It explains how this method can be beneficial for managing complex workflows and improving the mode…
-
json-repair excels at extracting LLM-wrapped JSON, outperforming jsonshim
A new benchmark, MALFORMED-300, evaluates how well parsers can extract JSON from malformed LLM outputs, particularly when JSON is wrapped in other formats like XML, HTML, or SQL. The json-repair tool achieved a perfect …
-
AI agents improve document authoring with structural conditioning, paper finds
A new paper explores the effectiveness of multi-agent systems in authoring formal documents, specifically focusing on tender responses. The research found that while converting documents to structural markup like XML im…
-
JSON parsers tested on LLM output: jsonrepair leads, JSON.parse fails
A comparison of three JavaScript JSON parsers—standard JSON.parse, jsonrepair 3.15.0, and JSON5 2.2.3—reveals significant differences in their ability to handle malformed LLM output. While JSON.parse correctly identifie…
-
Anthropic, OpenAI prompt engineering styles differ on delimiters
Prompt engineering guidance from Anthropic and OpenAI differs, with Anthropic recommending XML tags and OpenAI favoring Markdown headings. This distinction arises because language models do not parse prompts like code; …
-
AI text rewriting for DOCX faces structural and semantic hurdles
Developing AI tools to rewrite text within DOCX files presents unique challenges beyond simple text extraction and insertion. The complex XML structure of DOCX files, with elements like field codes, run-level formatting…
-
XML Prompting Enhances Small Language Model Performance
XML prompting is a technique designed to improve the performance of Small Language Models (SLMs) by providing structured context. This method uses a format similar to HTML to delineate instructions, variables, and examp…