#python #ai #data #data_structures #database #long_term_memory #machine_learning #ml #mlops #mongodb #pytorch #scikit_learn #sklearn #torch #transformers #vector_search

https://github.com/SuperDuperDB/superduperdb

#python #agent #ai #aiagent #application #chatbots #chatgpt #embeddings #llm #long_term_memory #memory #memory_management #python #rag #state_management #vector_database

Mem0 is a special tool that helps AI systems remember things. It makes AI interactions more personal and efficient by storing user preferences and past conversations. This means you don't have to repeat information, and the AI can give better answers based on what it knows about you. Mem0 also saves money by only sending important data to AI models, reducing costs up to 80%. It's easy to use and works with popular AI platforms like OpenAI and Claude.

https://github.com/mem0ai/mem0

#python #agent #ai #aiagent #awesome #chatgpt #hacktoberfest #hacktoberfest2025 #llm #long_short_term_memory #memori_ai #memory #memory_management #python #rag #state_management

Memori is an open-source memory engine that gives AI language models human-like memory using standard SQL databases like PostgreSQL, MySQL, or SQLite.[1][2] With just one line of code, you can enable any LLM to remember conversations, learn from interactions, and maintain context across sessions.[1] The key benefits are significant cost savings of 80-90% compared to expensive vector databases, complete data ownership and transparency since memories are stored in SQL databases you control, and zero vendor lock-in allowing you to export and move your data anywhere.[1][3] Memori works with popular frameworks like OpenAI, Anthropic, and LangChain, making it easy to integrate into existing projects without complex setup.[1]

https://github.com/GibsonAI/Memori

#python #large_language_models #llms #long_video_understanding #multi_modal_llms #rag #retrieval_augmented_generation

Vimo is a desktop app that lets me chat with any video, from short clips to hundreds of hours, in simple natural language. I can drag and drop videos, ask questions, find exact moments, compare multiple videos, and export useful insights, all on macOS, Windows, or Linux. Powering this is the VideoRAG algorithm, which deeply understands visual, audio, and contextual information, giving accurate answers even for very long videos. This helps me save time, understand complex content faster, and turn large video libraries into searchable, usable knowledge.

https://github.com/HKUDS/VideoRAG

#python #audio_driven_talking_face #dance_generation #end_to_end_filming #long_video_generation #video_diffusion_transformers

Stable Video Infinity (SVI) lets you create videos of any length—from seconds to minutes or infinite—starting from one image and text prompts, with no quality loss, drift, or repetition. It uses error-recycling to fix mistakes automatically, supporting stories, cartoons like Tom & Jerry (up to 10+ minutes), talking heads with audio, dancing from skeletons, and multi-scene films. Everything is open-source with models, scripts, datasets, and ComfyUI workflows on Hugging Face and GitHub—you can download, run demos, or train your own easily. This saves you time and effort for pro filmmaking or fun animations without limits.

https://github.com/vita-epfl/Stable-Video-Infinity

#python #infra #long #nvfp4 #parallel #real_time #video_generation

LongLive 2.0 helps generate long videos faster by using NVFP4, parallel processing, and multi-shot support for training and inference. It can reach up to 45.7 FPS while keeping good quality, and it also supports real-time, user-guided video generation. The benefit is that you can make long videos more quickly and efficiently, with less computing cost and smoother interactive use.

https://github.com/NVlabs/LongLive