nfield: schema-splitting for large-scale structured LLM extraction
nfield is a method that splits a large JSON schema into smaller sub-schemas and makes separate LLM calls for each, then merges the results. It solves the problem of frontier models scoring 0% on 369-field SEC filing extraction by achieving 85% accuracy with a 27B open model.
Entities
Related
Open-source 27B model on Groq achieves 85% on ExtractBench, frontier models score 0%
A 27B open-source model running on Groq scored 85% on a 369-field extraction task from SEC filings, while six frontier models scored 0%. The benchmark, ExtractBench (arXiv:2602.12247), uses real documents with human-checked answers.
Software engineer publishes final part of LLM-from-scratch series covering inference and decoding
A software engineer published the fourth and final part of a blog series explaining LLMs from the ground up, focusing on token-by-token generation, KV cache, and decoding strategies (temperature, top-k, top-p). The series aims to help other software engineers understand the internals of LLMs.
ddiff: LLM-generated diff format for isolating feature sets in code
ddiff is a format for isolating feature sets in code using LLM-generated diffs. It works by prompting an LLM to produce a diff of analysis of intents and code changes related to specific features, which can then be used by another LLM to implement the feature natively. The creator provides a live chat to Telegram group and a markdown WYSIWYG editor with rich uploads.
Upload project folder to get optimized markdown for LLM context
A web tool that lets users upload an entire project folder and receive a single, clean, optimized markdown file ready to paste into Claude or Codex. It solves the problem of manually preparing context for LLMs by automatically consolidating and formatting code files. All processing is done client-side for privacy.
Developer seeks feedback on fine-tuning LoRA for conversation state extraction in long LLM chats
A developer is working on a side project to improve AI conversation continuity by training a small model to extract structured conversation state from chat chunks, rather than relying on summarization. They are seeking feedback on their approach involving fine-tuning a LoRA, dataset design, and long-context systems.
