This repository is a comprehensive benchmarking suite for Rust serialization libraries. It compares over 20 serialization frameworks (e.g., bincode, capnp, flatbuffers, rkyv, prost, serde_json) across four synthetic datasets (log, mesh, minecraft_savedata, mk48) that model realistic data shapes. Benchmarks measure serialization/deserialization speed, payload size, compression (zlib, zstd), and zero-copy access. The codebase includes a deterministic data generation framework via the `Generate` trait, a tooling pipeline (bencher, parser, formatter) for running benchmarks and auto-updating the README with results, and an interactive Yew-based web page for visualizing throughput under bandwidth/CPU constraints. While primarily a research/benchmarking project, it also demonstrates high-quality patterns for conditional compilation, feature-gated code generation, and shared schema definitions across CLI and web frontend.
This repository is a curated, version-controlled taxonomy of cryptocurrency and blockchain ecosystems and their associated GitHub repositories. It is the dataset behind Electric Capital's developer activity reports. The primary artifact is a large collection of timestamped migration files (a custom DSL using commands like `repadd`, `ecoadd`, `ecocon`) that incrementally build a graph mapping ecosystems to repositories with tags. The dataset spans over a decade (2014–2026) and includes hundreds of thousands of repository entries across dozens of ecosystems, from major chains (Ethereum, Bitcoin, Stellar) to niche projects and hackathon submissions. The repository also provides a Python CLI tool (`open-dev-data`) for validating migrations, exporting the taxonomy as JSON Lines or Parquet, downloading the full dataset with resumable transfers and integrity checks, and exploring it locally via DuckDB and an interactive TUI (Harlequin). The codebase is well-structured, with clean abstractions for the data model, timestamp handling, and shell-like command parsing. However, the dataset itself suffers from significant noise—many entries are student hackathon repos, boilerplate forks, or duplicate URLs—which dilutes its signal for research. The curation is manual and not automatically deduplicated. Despite this, the taxonomy and tooling are highly reusable for any domain needing a version-controlled, hierarchically categorized registry of external projects.
This repository is a curated, comprehensive geographic database of countries, states, cities, and postcodes, with a strong emphasis on ISO 3166 compliance and data quality. It is primarily a data_research project, though it includes supporting application code (export pipelines, validation tools, CI/CD) that enables multi-format distribution (JSON, CSV, SQL, Parquet, etc.). The raw data lives in the `contributions/` directory as JSON files, which are the source of truth. A rich ecosystem of Python, PHP, and Node.js scripts handles ingestion from community and official sources, idempotent correction, enrichment (timezones, translations), and export to 12+ formats. The repository is actively maintained by a community of contributors and automated bots, with rigorous PR validation (schema, coordinates, duplicates, cross-references) and extensive fix documentation. It answers questions like 'What are the administrative divisions of a country?' and 'What postcodes exist for a given locality?' and is used as a reference data source for applications worldwide.
This repository is the canonical metadata registry for the Initia blockchain ecosystem. It contains structured JSON files describing chains (Cosmos SDK L1 and L2 minitias), their asset lists with full provenance traces (IBC, OP-bridge, wrapped), profile metadata for ecosystem projects, and error codes for Move smart contracts. The data is validated against JSON Schema (chain.schema.json, assetlist.schema.json, profile.schema.json) via GitHub Actions, ensuring consistency and correctness. The registry also includes TypeScript packages that export aggregated data and auto-generated Zod schemas for runtime validation, as well as an API build system that transforms and deploys the data to a CDN. While the core is data, there are significant application components: npm packages, validation workflows, and utility scripts for syncing images and computing IBC denoms. The repository solves the problem of discovering chains, assets, and bridge routes within the Initia network, serving wallets, explorers, and frontends.
This repository is a comprehensive collection of JavaScript interview questions and answers, serving as an educational resource for developers preparing for technical interviews. It consists of hundreds of questions organized by topic and difficulty, each with detailed explanations, code examples, and further reading references. The content is structured with metadata enabling categorization, filtering, and multilingual support (English, Portuguese, Chinese). The repository includes automated systems for generating the README from structured data, validating content quality, and managing translations. As a public mirror of GreatFrontEnd's main repository, it provides a searchable, well-organized knowledge base covering core JavaScript concepts, web APIs, security, performance optimization, testing, and design patterns.
Autoresearch is a self-contained autonomous ML experimentation framework created by Andrej Karpathy. Its core premise is that an AI agent (e.g., Claude or Codex) modifies a single training script (`train.py`), runs a fixed 5-minute training experiment on a single GPU, evaluates validation bits-per-byte (BPB), and keeps or discards changes based on whether the metric improved — all without human intervention. The human programs the agent via `program.md`, which serves as a lightweight instruction set for the experiment loop. The codebase has two fixed files and one editable file. `prepare.py` handles data downloading (parquet shards from HuggingFace), BPE tokenizer training (via rustbpe), a custom BOS-aligned packed dataloader with best-fit document packing (zero padding), and a fixed BPB evaluation harness. `train.py` contains a GPT model implementation with RoPE, sliding window attention (SSSL pattern), ResFormer value embeddings, RMS normalization, and a hybrid Muon+AdamW optimizer with per-parameter-type learning rate scheduling. The model uses `torch.compile` and runs in bf16 with Flash Attention 3. The `analysis.ipynb` notebook provides post-hoc visualization of experiment results from `results.tsv`. The project is explicitly designed for a single NVIDIA GPU (H100 tested), requires no distributed training infrastructure, and pins all dependencies including PyTorch 2.9.1 with CUDA 12.8. The architecture is deliberately minimal — the entire research loop runs in two Python files with roughly 1,600 lines of non-comment code total.
This is the official repository for the TidyTuesday project, a weekly social data project that provides curated datasets for practicing data wrangling and visualization, primarily in R. The repository contains hundreds of datasets organized by year and week (from 2018 to 2026), each with accompanying metadata, cleaning scripts, and documentation. It also includes GitHub Actions workflows for automated validation and deployment, as well as a set of R scripts for dataset curation and quality control. The repo serves as a community resource for learning data science skills through hands-on practice with real-world data. While not a software application, it demonstrates strong data engineering practices in its curation pipeline and validation infrastructure.
This repository is a community-curated directory of public APIs, organized by category in a single README.md file. Each entry includes the API name, description, authentication type, HTTPS support, and CORS status. The primary data artifact is the markdown file itself, which is manually maintained through pull requests. Supporting Python scripts validate the markdown format (alphabetical order, column padding, description length) and check that all listed API links are reachable, with special handling for Cloudflare-protected endpoints. The validation scripts are well-structured, parameterized, and include a comprehensive test suite, making them reusable for any similar markdown-based table validation. The repository is actively maintained with frequent commits and CI workflows, though its Python dependencies are slightly outdated (certifi 2021, requests 2.27.1). The core value is the curated dataset, not a software application.
This repository is a comprehensive educational curriculum for Machine Learning, structured as a collection of Jupyter notebooks, data files, and instructional documentation. While it contains small application examples (a Vue.js quiz app and a Flask web app demo), the dominant character is data research and education, focusing on teaching ML concepts through practical data science workflows. The content covers Regression, Classification, Clustering, NLP, Time Series, and Reinforcement Learning, utilizing Scikit-learn as the primary modeling library. The codebase is organized into modular lesson directories, each containing notebooks, assignments, and solutions. It demonstrates a strong commitment to accessibility with extensive multi-language translations and a devcontainer setup for consistent environments. However, reproducibility is impacted by a reliance on external datasets (Kaggle, Dropbox) which are prone to link rot, and scattered dependency management across different lesson folders rather than a unified root configuration. Technically, the repository shows signs of medium drift, particularly in the quiz application which utilizes Vue 2 patterns instead of the modern Vue 3 Composition API. The core ML content remains relevant, but the educational infrastructure (translation services, external links) requires ongoing maintenance to ensure accuracy and availability. Despite this, it stands as a high-quality, reusable resource for learning ML pipelines and data engineering patterns.
This repository is a comprehensive educational curriculum for Data Science, structured as a 10-week, 20-lesson course designed for beginners. It functions primarily as a data_research asset, containing Jupyter notebooks, R scripts, Python examples, and curated datasets (CSV, TSV) used for Exploratory Data Analysis (EDA) and visualization. The content covers the full data lifecycle, from data definition and ethics to cloud deployment on Azure, utilizing tools like Pandas, Matplotlib, Seaborn, and ggplot2. While the core is educational data content, it includes a hybrid application component: a Vue.js-based quiz application (`quiz-app`) used for assessments. The repository is heavily localized, with content translated into dozens of languages using an AI translation service, which introduces a specific maintenance pattern involving disclaimers about translation accuracy. The data lineage is well-documented, sourcing datasets from Kaggle, USDA, and Wikipedia, with clear licensing information provided. The codebase is optimized for learning rather than production engineering. Notebooks serve as the primary execution environment, demonstrating data cleaning, visualization, and basic statistical analysis. The architecture is flat and modular, organized by lesson topics (Introduction, Working With Data, Visualization, Lifecycle, Cloud). It is highly valuable for educators and learners seeking reproducible data science examples, though the heavy reliance on AI translation and mixed Vue versions (2 vs 3) presents technical drift risks.
This repository is a curated directory of public APIs organized by category. It serves as a reference resource for developers seeking APIs for various functionalities, ranging from advertising and analytics to weather and videogames. Each API entry includes the service name, description, and indicators for open-source or trial-based availability. The primary value lies in the structured, categorized listing of APIs maintained through community contributions. The repository implements basic data quality controls via a Travis CI configuration that uses awesome_bot to validate links in the README, ensuring the directory remains accurate and functional. Contribution guidelines enforce consistent formatting, alphabetical sorting, and proper marking of API types. While not containing complex data processing pipelines, the repository represents a well-maintained dataset with clear governance. Its strength is in the comprehensiveness of the API catalog and the reliability maintained through automated link validation and strict contribution standards. The project follows a typical 'awesome list' pattern common in developer communities for resource discovery.
This repository is a comprehensive educational course titled 'AI Agents for Beginners', structured as a collection of Jupyter notebooks and code samples. It functions primarily as an ML notebook collection and tutorial suite rather than a deployable application or library. The content focuses on teaching the Microsoft Agent Framework (MAF) and Azure AI ecosystem, covering topics from basic agent concepts to advanced protocols like MCP and A2A. The codebase is organized into 18 lessons, each containing Python and .NET code samples, documentation, and images. It relies heavily on external cloud services (Azure AI Foundry, Azure OpenAI) and rapidly evolving AI standards, making it a high-quality learning resource but one with significant dependency volatility.
XIU2/TrackersListCollection is a curated, daily-updated collection of BitTorrent tracker URLs. It provides several pre-filtered lists (best, all, HTTP-only, no-HTTP) in both plain text and Aria2-compatible formats, served via CDN. The core of the repository is a GitHub Actions workflow (`update-trackers.yml`) that automatically fetches trackers from multiple public sources, normalizes and deduplicates them, applies a blacklist, validates DNS resolution, and performs TCP connectivity checks on HTTP/HTTPS trackers. The resulting lists are committed back to the repository and mirrored to Bitbucket. While the repo is essentially a static data repository with no application logic, the CI pipeline demonstrates a clean, modular ETL pattern using bash and standard Unix tools. The primary value is the curated tracker data itself, which is widely used in BitTorrent clients to improve download speeds.
This repository is a research codebase for training and evaluating draft models for speculative decoding, supporting three algorithms: DSpark, DFlash, and Eagle3. It provides a full pipeline from data preparation (downloading prompts, regenerating answers with a target model via SGLang, and precomputing a large binary cache of target hidden states) to distributed training with FSDP and BF16, and evaluation on standard LLM benchmarks (GSM8K, MATH500, AIME, HumanEval, etc.). The code is well-structured with reusable components for configuration, distributed utilities, checkpointing, metrics, and custom attention masks. While primarily a research artifact, it exhibits strong engineering practices and several modules that could be extracted for other large-scale ML projects. The repository is hybrid in nature: it contains both application-like training/evaluation scripts and data_research elements such as data pipelines and evaluation datasets.
Hermes Agent is a self-improving AI agent platform with a sophisticated data architecture centered around session management, memory systems, and tool execution pipelines. The primary data lineage flows from user interactions (CLI, TUI, messaging platforms) through a gateway layer that manages session state, tool execution, and model communication. Raw conversation data is ingested via multiple transports (stdio, WebSocket, HTTP) and persisted in SQLite databases (state.db for sessions, kanban.db for task orchestration). The system transforms conversations through context compression, memory extraction (via plugins like Hindsight, Mem0, Honcho), and skill-based procedural memory. Analytical schemas expose session analytics, token usage, model performance, and skill effectiveness through REST endpoints and a web dashboard. The repository answers research questions about autonomous agent behavior, context management strategies, and multi-modal tool orchestration patterns.
The LOLBAS project is a curated dataset of Windows binaries, scripts, and libraries that can be abused for malicious purposes (Living Off The Land Binaries and Scripts). Each entry is documented in YAML files with a standardized schema, including technical details, use cases, MITRE ATT&CK mappings, detection artifacts, and external references. The repository serves as a threat intelligence resource for offensive security research and defensive security operations, enabling threat hunting and detection rule development. Data is organized by file type (binaries, libraries, scripts) and maintained through community contributions with automated validation via Pydantic schemas.
This repository provides a Claude Code Skill for reducing AIGC detection rates in Chinese academic writing. The Skill is available in two versions: a compact SKILL.md (~414 lines) for efficient daily use and a full SKILL_full.md (~554 lines) with detailed examples for maximum effectiveness. It does not contain executable code but offers a comprehensive set of rewriting rules based on 16 identified AI writing patterns across content, language, and style layers. The core methodology involves pattern recognition, constrained rewriting (7 hard constraints), a 4-step SOP (risk scanning → rewriting → rhythm check → noise preservation), and a 6-dimension quality scoring system. The Skill transforms input text by breaking AI predictability while preserving academic meaning and injecting human-like variability through techniques like theoretical repositioning, asymmetric restructuring, and judicious noise retention.
This repository is a curated 'awesome list' of password cracking tools, research, and resources — not a software application. The primary artifact is a single, comprehensive README.md that catalogs ~150 tools and resources across 15+ categories: hashcat automation/distributed cracking/rules, John the Ripper tooling, cloud-based cracking (AWS/Colab), wordlist generators and analyzers, hash conversion utilities, ML/AI password guessing models, academic papers (2009-2026), and notable researchers in the field. The list is alphabetically sorted, CC0-licensed, and maintained by a human curator who explicitly rejects AI-generated contributions. The repo's engineering value lies in its community governance infrastructure. A GitHub Actions workflow (pr-quality.yaml) uses peakoss/anti-slop to automatically detect and close AI-generated or spam PRs, enforcing account age, profile completeness, PR size limits, and maintainer-can-modify permissions. A second workflow runs awesome-lint to enforce list formatting standards. The PR template requires AI self-disclosure via a magic string. This governance stack is a genuinely reusable pattern for any community-curated repository facing AI slop. The content is current (papers through 2026) and the repo is actively maintained with zero open issues. While it contains no application code, it serves as a high-value reference index for security researchers, penetration testers, and anyone working on password recovery or authentication security. The hybrid nature: it's a curated data resource (the list itself) wrapped in application-grade CI/CD governance.
This repository provides the infrastructure for the Anticipatory Music Transformer, a research project for music generation using autoregressive transformers with anticipatory infilling. The codebase covers the full pipeline: preprocessing of MIDI files (Lakh MIDI dataset) into a compound token representation, tokenization into arrival-time or interarrival-time sequences with optional augmentation for anticipation training, and sampling from trained models using custom generation logic. It also includes scripts for human evaluation reproduction and model benchmarking. While primarily a research codebase for a specific paper, the library components in the `anticipation` package (particularly the conversion, ops, and sampling modules) are well-structured and could be reused in other music tokenization or generation projects. The repository does not include the training loop itself but provides configuration files for use with the Levanter training framework.
This repository is a comprehensive survey paper on the intersection of knowledge graphs and multi-modal learning. It organizes the literature into two main directions: KG-driven multi-modal learning (KG4MM) and multi-modal knowledge graphs (MM4KG). The survey covers various tasks such as visual question answering, image classification, multi-modal named entity recognition, and knowledge graph completion, among others. Each section includes a list of relevant papers with links to their sources (e.g., arXiv, conference proceedings). The repository is maintained through updates that include adding new papers and fixing links, as evidenced by the recent commit in 2026 and the news items up to 2024-12.