AI Trending

Project directory

Projects

Open-source AI projects tracked by the leaderboard, ordered by GitHub stars.

Python 8 stars

romilandc/langchain-RAG

A RAG implementation on LangChain using Chroma vector db as storage. Take some pdfs, store them in the db, use LLM to inference.

Overview: A RAG implementation on LangChain using Chroma as storage, allowing for the indexing and inference of PDFs using a Large Language Model. Key features: - Integration with LangChain - Use of Chroma for vector database storage - Support for both test and user-provided PDFs - Utilization of a Large Language Model for inference Tech stack: Python Why it matters: This implementation provides a practical and efficient way to leverage large language models with structured data, making it suitable for applications requiring comprehensive and context-aware responses.
Python 8 stars

rxwpgd2t3zs2k/seedream-5-0-lite-api-ru

Сидрим — Seedream 5.0 Lite API (`seedream-5-0-lite`): model id, pricing, parameters and copy-paste examples for the OpenAI-compatible APIMart endpoint.

Overview: Seedream 5.0 Lite API is a comprehensive solution for generating high-quality images with advanced text rendering, offering a cost-effective and scalable model that can be easily integrated into any project. Key features: - Supports OpenAI-compatible API endpoint - Provides 4K image generation with up to 15 images per batch - Pricing is per delivered image, with a minimum cost of $0.028 per image - Supports batch processing and reference images - Asynchronous processing by default, with task status feedback - Compatible with various image and language models Tech stack: Python Why it matters: Seedream 5.0 Lite API simplifies the integration of advanced image generation capabilities into projects, offering a cost-effective and scalable solution that can be easily integrated into any project. Repository: rxwpgd2t3zs2k/seedream-5-0-lite-api-ru Primary language: Python GitHub stars: 8 GitHub description: Seedream 5.0 Lite API (`seedream-5-0-lite`): model id, pricing, parameters and copy-paste examples for the OpenAI-compatible APIMart endpoint.
TypeScript 8 stars

ryhkml/fine-tune-forge

JSONL generator designed for models like Google PaLM 2 and OpenAI GPT-3.5

Overview: FineTuneForge is a tool designed for generating JSON Lines (JSONL) to facilitate the fine-tuning of AI language models like Google's PaLM 2 and OpenAI's GPT-3.5. Key features: - JSONL generator - Easy transformation of text data into a JSONL format - Suitable for fine-tuning Google's PaLM 2 and OpenAI's GPT-3.5 models Tech stack: TypeScript Why it matters: FineTuneForge simplifies the process of preparing data for model fine-tuning, making it easier for developers to optimize AI language models for specific applications.
Python 8 stars

s78ve5v4x9ji2/gemini-3-pro-image-preview-api-id

Nano Banana Pro API (`gemini-3-pro-image-preview`): model id, pricing, parameters and copy-paste examples for the OpenAI-compatible APIMart endpoint.

### Overview: Nano Banana Pro API (`gemini-3-pro-image-preview`) is a model id, pricing, parameters, and copy-paste examples for the OpenAI-compatible APIMart endpoint. It offers a single key for all catalogs, compatible with 300+ models, and supports 4K image generation and editing with multi-step planning. ### Key Features: - **Model ID and Routes**: The API uses the model id `gemini-3-pro-image-preview` for 4K image generation and editing, with a pricing tier of $0.03 per image (4K $0.04). - **Pricing**: The pricing is per delivered image, with tiered costs based on resolution. The effective price is $0.03 per image. - **Request Parameters**: The API supports parameters such as `model`, `prompt`, `size`, `resolution`, `n`, `image_urls`, and `nsfw_check`. - **Asynchronous**: The API is asynchrnous from the start, requiring `task_id` polling for updates. ### Tech Stack: - Python ### Why It Matters: - **Single Key for All Catalogs**: A single API key covers all Nano Banana Pro models, making it easy to manage and scale. - **Minimum $1, Pay as You Go**: No need for a monthly subscription; pay only for usage. - **Cost Reporting**: The cost is reported in the response, allowing for accurate billing. - **Asynchronous Operation**: The API is designed to be asychronous,
Jupyter Notebook 8 stars

saadhaxxan/Sentiment-Analysis-with-BERT

Overview: Sentiment Analysis with BERT is a project that uses BERT (Bidirectional Encoder Representations from Transformers) for text classification tasks, specifically for sentiment analysis. Key features: - Implements BERT for sentiment analysis - Includes a Jupyter Notebook implementation - Provides a run-in-place option using Google Colab Tech stack: Python, Jupyter Notebook, BERT Why it matters: This project demonstrates the effectiveness of BERT in natural language processing tasks, particularly in sentiment analysis, which can be useful in various applications like customer feedback analysis, social media monitoring, and more.
Python 8 stars

santos-sanz/mlx-lora-finetune-template

Template for fine-tuning LLMs with LoRA using Apple MLX on Mac Silicon

Overview: This repository is a comprehensive template for fine-tuning Large Language Models (LLMs) using Low-Rank Adaptation (LoRA) and Apple MLX on Apple Silicon. It includes a user-friendly Streamlit UI for easy configuration and training, as well as features like checkpoint management, data preparation, and integration with Hugging Face. Key features: - Fine-tuning with LoRA - GRPO Reinforcement Learning - MLX Optimized for Apple Silicon - Streamlit UI for configuration and training - Hugging Face integration - Data preparation and checkpoint management - AI-assisted data preparation Tech stack: Python Why it matters: This template streamlines the process of fine-tuning LLMs, making it accessible for developers and researchers, especially on Apple Silicon hardware, where it can leverage the full potential of Apple Silicon's performance optimizations. Repository: GitHub Repository: santos-sanz/mlx-lora-finetune-template GitHub stars: 8 GitHub description: Template for fine-tuning LLMs with LoRA and MLX on Apple Silicon.
Python 8 stars

sebasmos/bodhi-llms

A lightweight, provider-agnostic wrapper that embeds epistemic virtues (curiosity and humility) into LLM responses

Overview: BODHI is a lightweight Python library designed to wrap any LLM and enhance its responses with curiosity, humility, and clarity. This makes the responses more genuine, specific, and contextually appropriate. Key features: - Enhances LLM responses with curiosity - Teaches LLMs to ask clarifying questions instead of making assumptions - Provides specific and clinically relevant advice - Addresses the problem of overconfidence and vague advice Tech stack: Python Why it matters: BODHI addresses the limitations of current LLMs by making them more human-like and contextually relevant, thereby improving patient outcomes and healthcare communication. Repository: sebasmos/bodhi-llms GitHub stars: 7 GitHub description: A lightweight, provider-agnostic wrapper that embeds epistemic virtues (curiosity and humility) into LLM responses. [PyPI](https://pypi.org/project/bodhi-llm/) [Python 3.10+](https://www.python.org/downloads/) [License: CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) [PLOS Digital Health](https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001013) [The Lancet](https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(25)01626-5/fulltext) [
Python 8 stars

sfpxvicw/midjourney-api-ru

Миджорни / Мидджорни — Midjourney API (`midjourney`): model id, pricing, parameters and copy-paste examples for the OpenAI-compatible APIMart endpoint.

### Overview: Midjourney API (`midjourney`) is an OpenAI-compatible API that provides access to a wide range of image generation models, including Midjourney, through the APIMart endpoint. This API allows developers to create high-quality images, videos, and text-to-image combinations without the need for a monthly subscription. ### Key Features: - **Model Support**: Supports various image generation models, including Midjourney. - **Billing**: Prices are per delivered image, with a base price of $0.045 per image. - **API Endpoint**: `POST https://api.apimart.ai/v1/images/generations`, with tasks being polled via `GET /v1/tasks/{id}`. - **Pricing Model**: Prices are based on the output type, with default being $0.045 per image. - **Model ID**: Supports multiple models, including Midjourney (model id: `midjourney`). ### Tech Stack: - Python ### Why It Matters: - **One Key Access**: A single API key provides access to the entire catalog of models, simplifying the integration process. - **Minimal Cost**: The API is priced on a per-image basis, with a minimal $1 charge for each task. - **Automatic Billing**: The API automatically returns the cost of completed tasks, ensuring transparency and ease of use. - **Asynchronous Processing**: Tasks are processed asynchronously, making the API suitable for various use cases without requiring additional integration. ### Repository: - **Repository**: sfpxvicw/midjourney-api
Jupyter Notebook 8 stars

simranjeet97/AI_Engineer_Roadmap_2027

A 10-week, staff-level curriculum and engineering roadmap focused on LLM inference systems, serving physics, custom kernels, disaggregated architectures, and production optimization.

### Overview: This 10-week staff-level curriculum and engineering roadmap is designed for AI Engineers, ML Systems Engineers, and Infrastructure Architects transitioning to Staff-level LLM Inference & Serving Systems. The curriculum focuses on the foundational principles of LLM inference systems, including hardware-bound physics, deep dive into PagedAttention and modern schedulers, production observability, advanced batching mechanics, prefix caching, multi-turn workloads, quantization tradeoffs, KV cache eviction, compression, speculative decoding, disaggregated prefill & decode serving, Kubernetes orchestration, queue-based autoscaling, model routing, token budgeting, and the public capstone benchmark. ### Key Features: - **Hardware Bound Physics**: Understanding the physical limits of GPU hardware and memory. - **PagedAttention & Modern Schedulers**: Deep dive into advanced attention mechanisms and modern scheduling techniques. - **Production Observability**: Techniques for monitoring and optimizing inference performance. - **Advanced Batching Mechanics**: Handling large concurrency scenarios to minimize latency. - **Prefix Caching & Multi-Turn Workloads**: Optimizing prefix caching and multi-turn workloads. - **Quantization Tradeoffs**: Balancing kernel performance and numerical accuracy. - **KV Cache Eviction & Compression**: Managing cache eviction and compression to optimize memory usage. - **Disaggregated Prefill & Decode Serving**: Implementing disaggregated prefill and decode serving architectures. - **Kubernetes Orchestration & Autoscaling**: Using Kubernetes for orchestration and queue-based autoscaling. - **Model Routing & Token Budgeting**: Optimizing model routing and token budgeting for efficient inference.
Go 8 stars

skorfmann/nanobanana

Lightweight CLI for AI image generation using Google Gemini API. Text-to-image, image editing, multi-image composition. Perfect for coding agents like Claude Code.

Overview: Nanobanana is a lightweight CLI tool designed for generating and editing images using Google's Gemini API, perfect for coding agents like Claude Code. Key features: - Text-to-image generation - Image editing - Multi-image composition - Model selection via `-model` flag - Flexible output options Tech stack: Go Why it matters: Ideal for automating image generation workflows with AI coding assistants, enhancing productivity and creativity in software development. Repository: skorfmann/nanobanana GitHub stars: 8 GitHub description: Lightweight CLI for AI image generation using Google Gemini API. Perfect for coding agents like Claude Code.
Python 8 stars

slinusc/bench360

Bench360 is a modular benchmarking suite for local LLM deployments. It offers a full-stack, extensible pipeline to evaluate the latency, throughput, quality, and cost of LLM inference on consumer and enterprise GPUs. Bench360 supports flexible backends, tasks and scenarios, enabling fair and reproducible comparisons for researchers & practitioners.

Overview: Bench360 is a comprehensive benchmarking suite designed to evaluate the performance, energy consumption, and quality of Local Large Language Models (LLMs) on consumer and enterprise GPUs. It provides a flexible pipeline for researchers and practitioners to analyze various metrics across different quantization formats, model sizes, and deployment scenarios. Key features: - Evaluates latency, throughput, quality, and cost for tasks like summarization, QA, and SQL generation. - Supports multiple quantization formats (FP16, INT8, INT4) and model sizes. - Simulates different deployment scenarios (single-stream, offline batch, multi-user server) to test inference engines. - Measures GPU/CPU utilization, energy consumption, and quality metrics like F1, ROUGE, and AST match. - Includes a workload controller for testing different inference engines under identical conditions. Tech stack: Python Why it matters: Bench360 helps researchers and practitioners make informed decisions about LLM deployment by providing a detailed analysis of performance, energy efficiency, and task quality across various parameters. This tool is essential for optimizing local LLM deployments and ensuring fair and reproducible comparisons. Repository: slinusc/bench360 GitHub stars: 8 GitHub description: Bench360 is a modular benchmarking framework for evaluating local LLM deployments across backends, quantization formats, model architectures, and deployment scenarios. It enables researchers and practitioners to analyze latency, throughput, quality, efficiency, and cost in real-world tasks like summarization, QA, and SQL generation under both consumer and data center conditions.
Swift 8 stars

somya-droid/Pirate-LLM-Server

Run local LLM servers on iPhone with OpenAI-compatible API, Metal GPU acceleration, and easy GGUF model management without cloud or jailbreak.

Overview: This project allows developers to run local LLM servers on their iPhone using OpenAI-compatible API, Metal GPU acceleration, and easy management of GGUF models without the need for cloud services or jailbreak. Key features: - Local LLM server - OpenAI-compatible API integration - Metal GPU acceleration - GGUF model management Tech stack: Swift Why it matters: This project offers a seamless, efficient solution for developers looking to integrate large language models into their applications without the need for cloud resources or the complexity of jailbreaking their devices. The use of Metal GPU acceleration ensures optimal performance, and the GGUF model management feature simplifies the process of managing and updating models on the device.
Python 8 stars

sriramsme/pragYantra

software for simulating human-like interaction with AI using a cognitive approach

Overview: PragYantra is a software designed to simulate human-like interaction with AI, focusing on embodied assistant capabilities such as seeing, hearing, and conversing. It is a modernization of a personal prototype written in Python with explicit application and device interfaces. Key features: - Embodied Assistant: Seeing, Hearing, Conversing, Memory, and a Boundary for Physical Sensors. - Typed Text and Transcribed Audio for Conversations. - Background WebCam Descriptions and Timestamped Sensor Readings for Context. - Synthesizing Spoken Replies. - Explicit Notes for Persistent Keyword Recall and Optional Semantic Recall. - Explicit Memory Commands Running in Application Code. - Explicit Names/Preferences Supplementing Notes. - Direct Image Inspections Returning Bounded Predicates, Choices, or Scores Independently of Generated Conversation. - Automated Integration/Lifecycle Tests and Live Hosted-Service Smoke Checks. - Microphone/Speaker Operation and Raspberry Pi Deployment Still Need Validation on Actual Devices. Tech stack: Python. Why it matters: PragYantra is a modernization of a personal prototype, designed to simulate human-like interaction with AI, focusing on embodied assistant capabilities such as seeing, hearing, and conversing. It is a modernization of a personal prototype written in Python with explicit application and device interfaces. GitHub stars: 8 Repository: sriramsme/pragYantra
C++ 8 stars

starkdg/mvptree

multiple vantage point distance-based tree data structure

Overview: MVPTree is a distance-based data structure designed for indexing multi-dimensional vector objects. It efficiently indexes data based on its distance from selected vantage points. Key features: - Multiple vantage points chosen directly from the indexed data - Distance-based indexing for efficient query performance Tech stack: C++ Why it matters: MVPTree offers a fast and effective way to index multi-dimensional data, particularly useful in applications requiring high-speed spatial queries.
TypeScript 8 stars

steve-aldrin/vecstore-js

A simple, fast, and pluggable vector storage for the browser

### Overview: VecStore-js is a fast, pluggable vector storage solution for the browser, perfect for offline-first AI features, browser extensions, and web apps that prioritize data privacy. ### Key features: - 🧠 Local Semantic Search: No server calls needed, all processing happens in the browser. - 🚀 High Performance: HNSW (Hierarchical Navigable Small World) for fast, approximate nearest-neighbor search, implemented in WebAssembly. - 🔌 Pluggable Components: Easily switch between search algorithms (HNSW, Cosine Similarity) and embedding models. - 🔒 Privacy-First: User data never leaves the browser. - Offline-First: Caches models and data for use without an internet connection. - Lightweight: Small footprint, designed for the client-side. ### Tech stack: TypeScript ### Why it matters: VecStore-js simplifies local semantic search, making it ideal for applications requiring privacy and offline-first functionality.
Python 8 stars

steveimm/comfyui-lance-aio

Run ByteDance Lance-3B (unified image/video generation, editing, and understanding) in ComfyUI, with dynamic VRAM streaming for low-VRAM GPUs

### Overview: This repository implements ComfyUI custom nodes to run Lance-3B model using ComfyUI, focusing on tasks such as text-to-image, text-to-video, image-text to video, image editing, video editing, and image understanding. It supports low-VRAM GPUs and is compatible with limited VRAM machines. ### Key Features: - Text-to-Image - Text-to-Video - Image-Text to Video - Image Editing - Video Editing - Image Understanding - Video Understanding ### Tech Stack: Python ### Why It Matters: This implementation allows for efficient use of low-VRAM GPUs, making it suitable for a wide range of devices with limited computational resources. It leverages ComfyUI's capabilities to provide a seamless user experience while maintaining high performance.
Python 8 stars

streamweights/streamweights

Build a small model for your task, on hardware you control. Bring labeled examples. Fine-tune locally, compare against simple baselines, and export to GGUF or safetensors. Pause and resume supported training runs across Mac and Linux.

Overview: Streamweights is a tool designed for developers that allows them to build small models for specific tasks on hardware they control. This approach enables them to fine-tune models locally and compare their performance against simple baselines, ensuring that the model is worth building. Key features: - Training a small model from labeled examples - Measuring the model's performance against simpler approaches - Exporting the model in GGUF or safetensors format - Support for Mac and Linux platforms - Ability to pause and resume training runs - Data stays local by default, with remote storage used when configured Tech stack: Python Why it matters: Streamweights simplifies the process of building and evaluating models, making it accessible for developers without requiring extensive machine learning expertise. It allows for fine-tuning models locally, ensuring that the model's performance is accurate and relevant to the task at hand. Repository: streamweights/streamweights GitHub stars: 8 GitHub description: Streamweights is a tool designed for developers that allows them to build small models for specific tasks on hardware they control. This approach enables them to fine-tune models locally and compare their performance against simple baselines, ensuring that the model is worth building.
TypeScript 8 stars

sujal12344/AI-Companion-Builder

AI-Companion is a cool software that lets you create your own custom AI models of people you admire, like actors or celebrities. It's a tool to make personalized artificial intelligence companions based on your favorite individuals.

Overview: AI-Companion is a software tool that allows users to create their own AI companions based on their favorite celebrities or actors. It provides a dynamic SaaS application built on Next.js 14, TypeScript, and other advanced technologies. The platform offers features like engaging with pre-existing AI models, creating custom models, managing models, and searching for existing models. Key features: - Engage with pre-existing AI models of celebrities - Create and customize your own AI models - Manage and update custom AI models - Search for existing AI models by name or category Tech stack: Next.js, Node.js, TypeScript, TailwindCSS, Stripe, Zod, Prisma, React Hook Form, Shadcn, zustand, OpenAI, Pinecone-database, Upstash, LangChain, Replicate Why it matters: AI-Companion empowers users to create personalized AI companions that can provide insights and engage them in meaningful conversations, making the experience of interacting with AI more personalized and engaging.
Rust 8 stars

sunormesky-max/epicode

Epicode - AI Memory Operating System. Spatial AI memory with knowledge graph, MCP integration, and auto-defense.

### Overview: Epicode is an open-source AI memory service designed for AI agents, providing spatial memory, semantic search, knowledge graph, temporal validity, MCP integration, and more. ### Key Features: - **Spatial Placement** - Memories are placed as tetrahedrons for clustering and visualization. - **Semantic Search** - Hybrid search using BM25 and HNSW for natural-language retrieval. - **Knowledge Graph** - Automatic relationship extraction and dynamic graph updates. - **Temporal Validity** - Supports `valid_from` / `valid_until` windows and freshness scoring. - **MCP Integration** - Discover the current tool catalog with MCP `tools/list`. - **L0 Active Inference** - Evidence-backed drive signals with REST/MCP inbox and acknowledgement flows. - **SMRP Protocol** - Transport-independent structured memory responses with tier, retrieval provenance, and optional topology/placement metadata. - **Multi-tenant Cloud** - User management, quotas, invite codes, and cloud deployment. ### Tech Stack: Rust ### Why it Matters: Epicode provides a robust and scalable solution for AI agents, enhancing their ability to store, retrieve, and manage spatial and temporal memories, making them more effective in various applications. ### Repository: sunormesky-max/epicode ### GitHub Stars: 8
Dockerfile 8 stars

suriya-mca/Docker_Compose_Files

In this repo, we have different docker-compose files for open-source and popular software and frameworks

Overview: This repository contains Docker Compose files for various open-source and popular software and frameworks, designed to set up these software or frameworks in a containerized environment. Key features: - Docker Compose files for SQL databases (SQLite3, PostgreSQL, CockroachDB, MySQL, MSSQL, Mariadb, TimescaleDB, EdgeDB) - Docker Compose files for NoSQL databases (Redis, MongoDB, KeyDB, Cassandra, CouchDB, Couchbase, Neo4j, Memcached, Valkey) - Docker Compose files for Vector databases (Qdrant, ChromaDB, PostgresVector, Weaviate) - Docker Compose files for Analytics databases (ClickHouse, Apache Pinot, Apache Kylin, Apache Druid) - Docker Compose files for DB clients (Dbeaver, Dbgate, Pgadmin4, Phpmyadmin, Beekeeper Studio) - Docker Compose files for Backend frameworks (FastAPI, Django, Spring Boot) - Docker Compose files for CRM/ERP (Odoo, ERPNext) - Docker Compose files for CMS (WordPress, Directus, Ghost, Grav, Strapi) - Docker Compose files for Docker clients (Portainer, Dokku) - Docker Compose files for Backend as a Service (Pocketbase, Appwrite) - Docker Compose files for API Platform (Hoppscotch) Tech stack: Dockerfile Why it matters: These Docker Compose files provide a standardized way to set up and manage open-source software and frameworks in a containerized environment, making it easier to deploy