Star 总榜

Carter is a streamlined platform designed to help users effortlessly organize, share, and manage their links. With a secure interface, it allows users to save, categorize, and retrieve links in a controlled environment. Emphasizing privacy, Carter keeps links accessible and shareable as needed, empowering users with full control.

The Ultimate Vectors Library for Java

Search on your images via text or image to image search. Uses OpenAI CLIP embedding and LanceDB

A RAG implementation on Llama Index using Qdrant vector stores as storage. Take some pdfs, store them in the db, use LLM to inference.

A terminal UI for browsing and inspecting vector databases

ToucanDB is a brand-new micro ML-first database engine 🦜

The notebooks & projects code for Generative for Backend Dev online bootcamp

An advanced AI-powered conversational agent leveraging the Llama 3.2 model and Phidata framework. Features include reasoning, natural language interaction, and tool integration for web searches and calculations. Designed for interactivity with enhanced logging, support for custom tools, and structured outputs.

Give non-multimodal Claude Code main models the ability to see pasted screenshots — a ~200-line UserPromptSubmit hook.

CLI & async Python library for free AI chat, image & video generation.

Characterize Anything: A Wondrous Chemical Reaction between vision models and AI Characters

SemEval 2022 Task 5: Multimedia Automatic Misogyny Identification - baseline models and dataset

[EMNLP 2025 Findings] MEXA: Towards General Multimodal Reasoning with Dynamic Multi-Expert Aggregation

ACL 2024 (SRW), Official Codebase of our Paper: "MoExtend: Tuning New Experts for Modality and Task Extension"

API/MCP wrapper for RagAnything

VideoEval-Pro: Robust and Realistic Long Video Understanding Evaluation [TMLR26]

A curated list of resources for Grok, the AI developed by xAI

[NAACL 2025 Findings] Mitigating Hallucinations in Large Vision-Language Models via Summary-Guided Decoding

[ICLR 2026] Official repository of "Uni-DPO: A Unified Paradigm for Dynamic Preference Optimization of LLMs".

Recurrence Meets Transformers for Universal Multimodal Retrieval