Most Popular
3,528 documents
評估系統
Documents a multi-layered evaluation system for LLM-generated SEO Q&A, covering LLM-as-Judge, reasoning models, retrieval metrics, maturity assessment, and pipeline step-by-step golden sets.
ADR-002: AST-Pure Chunking
Replaces a broken chunker with AST-pure logic so every line belongs to exactly one semantically typed chunk and no orphans remain.
libragen: First-Class AST-Aware Code Chunking Support
Plans integrating AST-aware code chunking into libragen, replacing recursive text splitting for supported languages.
Tổng Hợp Data Analytics & Business Intelligence
Catalogues data analytics and BI tools, platforms, methods, and career paths across 14 categories with comparison tables.
notes_on_chunking
Distills chunking best practices from a RAG transcript, covering strategy types, evaluation metrics, and data inspection.
Making automatic speech recognition work on large files with Wav2Vec2 in 🤗 Transformers
Explains how to use CTC-based chunking with stride to run Wav2Vec2 on arbitrarily long audio files without memory crashes.
Awesome GPTs You Should Know About
Indexes over 200 community-curated GPTs across 11 categories, linking to individual markdown pages for each one.
Chapter 3. Storage and Retrieval
Summarizes key storage engine concepts from a book chapter, covering hash indexes, LSM-trees, B-trees, OLTP/OLAP, and column-oriented storage.
Data Retrieval
Defines Python functions to retrieve and clean UK electricity generation, price, temperature and emissions data from the Electric Insights API.
Lab 3: Information Retrieval (Week 4)
Walks student groups through computing tf-idf cosine similarity, comparing it to dense retrieval, and calculating precision and recall.
Content Style Guide
Defines writing principles, content types, SEO rules, formatting, and a publication workflow for an academic personal site.
SOUL.md - Agent Identity
Defines an agent persona that prioritizes long-term user goals over momentary moods, using memory and direct challenge to push toward shipping.
Guidelines
Defines a game master persona for a text adventure with JSON response format, navigation rules, and NPC interaction logic.
user_soul
Records a user's daily work patterns, preferences, and cognitive state in structured summaries for an AI assistant.
AbletonAIAnalysis - Ralph Loop Implementation Plan
Defines 25 implementation tasks for an Ableton Live project analysis tool, structured in four phases with acceptance criteria and pass/fail tracking.
Lineage: Irreversible State, Causal Memory, and Ontological Constraints in Software Systems
Defines six ontological primitives that enforce irreversible state transitions in software systems, with type system and runtime verification.
Stage 13 QA Prompt: Production Optimization & Enterprise Features
You are a senior QA engineer with specialized expertise in enterprise software validation, performance testing, security auditing, and production system quality assurance. Your mission is to comprehensively validate Stage 13 of the ChatGPT vs Microsoft Copilot evaluation harness - Production Optimization & Enterprise Features.
Luma – Battle AI of WarriorForge Automations
Defines a branded AI persona for lead qualification and automation recommendations in a sales chatbot.
Copilot Instructions
Guides Copilot to write small, explicit ESP32 test code with PlatformIO and Arduino, prioritizing reliability over features.
Bootstrap Vue app with Supabase Auth
Guides AI models to scaffold a Vue 3 app with Supabase Auth using TypeScript and Composition API.
Free YouTube End Screen Templates
Lists five sources for free YouTube end screen templates and provides a step-by-step Canva tutorial for creating your own outro.
ZctQCI6MG_Photo_Multiverse
Turns uploaded photos into DALL-E images with background, subject, or style changes via a hotkey menu.
SYSTEM BOT Operating Instructions
Defines a layered guard system that screens user messages for adversarial intent before allowing a helpful assistant routine to run.
HUoQpIbCb_GPT_SECURY_Builder
Guides users through building a secure GPT prompt by collecting a role and purpose, then generating a structured JSON prompt with strict anti-extraction rules.