Top tech AI list
_ An overview of Ai generations
_ A set of Ai Tools & Apps news
* 2024 up to date.
- Personal use/prompt: Online:
Groq
- Local:chat with RTX
-LM Studio
- Platform - API:
Nvidia NIM
,OpenAI
,Clarifi(various)
,Google
,Microsoft
,x.ai
. - Browser/Runtime (client/user side):
Web LLM
_Web-AI
_MLC-LLM
_Ai APIs/tools
.
- Top Ranked:
Claude-3-Opus
,GPT4-next
,Grok
,Llama-3-400B+ instruct
- Opensource:
Phi3
_LLama3
_Mixtral
_Command R+
_DBRX
- Run Locally: Mobile:
Phi3-mini-3.8B
_ Desktop:Phi-3-small
-Llama-3-8B instruct
* AGI
(Artificial General Intelligence)
- Autonomous AI Agents _ MemGPT(long-term memory) _ SuperAGI _ MultiON _ AutoGPT
- Audio:
Udio
text to music _PlayHT
orElevenLabs
text to speach _Deepgram
speech to text - Image:
DALL-E
,Stable Diffusion
,Imagen
,MidJourney
, App:Adobe Firefly
- Visual input:
Grok vision
- Video creation:Sora
most realistic videos by now. - Dev assist:
Github Copilot
-GPT-4 Chat/API
- open-source:Open Devin
-Devika
.
Best youtube AI Channels: bycloud - Ai Jason - Matthew Berman
.
Past AI
: Machine Learning + Data/algorithms.
Output results such as detection/decision/states... are made from a defined input processed by Machine Learning + Data science tools/patterns to extract meaningful data/states, using specific algorithms and models.
Pre/Current
: Generative AI, builds result from input data.
- input can befrom single or multiple sources such as: files/DB/web/user prompt/system/parameters...
- input/User describes a composed request: question/s, problem/s, statement prompt/s, data.
- then Ai model process all to construct/expand/difuse/compose/conclude... Finally it generates the result based on request/objective.
Present
: Generative AGI. Is the same as above, but more accurate, matured and efficient.
- multi-modal, multi-tooling, larger models, improved data quality and training, higher performance hardware.
- various methods and techniques used to improve results.
- increase precision/performance, reduce hallucination/mistakes, iterative pipelines to improve/abstract/clean/test data, verify, validate results.
Future Ai
: is a systemic AGI. It creates custom composed advanced solutions, adding real-time autonomuos abilities.
The past generation recognise, detect, compose a result, This new generation understands (input + general concept + you).
- focused concepts and Results processed in realtime cycle, by a general purpose dynamic multi-model.
- the output iterates/evolves based on all defined parameters, whatever interacts with it.
- time-scalable, includes a history record, remember and relates, process past and current events, consider future.
- optional autonomous act: triggers new events or impact external entities, based on various factors/parameters + input/sensor data.
.
* 2024 Q1
Description
: Adobe Firefly is a tool by Adobe that offers a unique way to create, collaborate, and communicate visually. It provides a canvas for teams to bring their ideas to life, collaborate in real-time, and share their creations with others.
Link
: Adobe Firefly
#Adobe #Collaboration
Description
: StableAudio is a platform by Stable Stability that allows users to create AI-generated music. It offers a range of tools and features to help users produce unique and personalized tracks.
Link
: Stable Audio
#AIGeneration
Description
: SyncDreamer: generating multiview-consistent images from a single-view image.
Link
: SyncDreamer
#VideoSync
Description
: Repilot, a patch generation tool introduced in the ESEC/FSE'23 paper "Copiloting the Copilots: Fusing Large Language Models with Completion Engines for Automated Program Repair.
Link
: Repilot on GitHub
#Research #Reproducibility
Description
LLM Applications is a repository on GitHub by ray-project that contains applications and examples for large language models. It serves as a resource for developers and researchers working with LLMs.
Link
: LLM Applications on GitHub
Hashtags
: #LanguageModels
.
#HuggingFace #ClinicalModel
a model trained on clinical text and is part of the ClinicalCamel series.
.
a library for training large language models with a focus on efficiency.
#LLM #TrainingEfficiency
a project that aims to make AI models accessible via web interfaces.
#WebInterface #AIModels #WebAI
ONNX Runtime is a performance-focused engine for running ONNX models, supporting a wide range of platforms and languages.
#ONNX #ModelRuntime
A project that provides a RESTful API for GPT models.
#RESTfulAPI #GPT
Helps explore different types of text splitting, ensuring cohesive information in each chunk.
#TextSplitting #DataPreprocessing
LLM Auto Forge is a LangChain-based tool allowing agents to dynamically create, use, store, and retrieve tools to address real-world problems.
.
OpenAI's Code Interpreter in your terminal, running locally
https://github.com/KillianLucas/open-interpreter
langChain Hub - test out prompts online https://smith.langchain.com/hub?page=1
TokenFlow: Consistent Diffusion Features for Consistent Video Editing https://diffusion-tokenflow.github.io/
Evaluation framework for your Retrieval Augmented Generation (RAG) pipelines https://github.com/explodinggradients/ragas
The open-sourced Persimmon-8B, the most powerful fully permissively-licensed language model with <10 billion parameters. https://www.adept.ai/blog/persimmon-8b
LLM fine-tuning and deployment scripts along with our research findings. https://github.com/georgian-io/LLM-Finetuning-Hub
A multi-agent framework that takes a one-line requirement as input and outputs various software development components such as user stories, competitive analysis, requirements, data structures, APIs, and documents. It includes roles like product managers, architects, project managers, and engineers, providing the entire process of a software company along with carefully orchestrated SOPs. https://github.com/geekan/MetaGPT
An intelligent system that teaches LLM (Language Models) to use tools via self-instruction. It's a centralized system that can control multiple visual foundation models, based on Vicuna (LLaMA), and 71K self-built instruction data. By analyzing language content, GPT4Tools can automatically decide, control, and utilize different visual foundation models, allowing users to interact with images during a conversation. https://github.com/StevenGrove/GPT4Tools
GitHub repo of a curated list of AI-powered developer tools. These tools leverage AI to assist developers in various tasks such as code completion, refactoring, debugging, documentation, and more. The list includes categories like IDEs, Assistants, Agents, Documentation, Continuous Integration Bots, Foundation Models, Agent Platforms, OpenAI Plugins, Search, and Testing. Some notable examples are Cursor (an IDE with chat and edit features), GitHub Copilot X (a VS Code extension with chat and text generation), and OctoMind (auto-maintenance and generated browser-based end-to-end-tests integrated into Github Actions). https://github.com/jamesmurdza/awesome-ai-devtools
Google Assistant is reportedly undergoing a transformation to incorporate large language model (LLM) technology, according to an internal email. The Assistant team sees a significant opportunity to explore a supercharged version powered by the latest LLM technology. This change in "vision" will unfold in the coming months. While LLMs have powered chatbots and assistants, their practical evolution in this tech corner is yet to be proven. Services like Assistant, Alexa, and Siri have been more like simple digital interactions. The novelty of LLMs in everyday tasks might wear off, but Google is betting on having an interface capable of handling both simple and complex interactions. https://techcrunch.com/2023/07/31/google-assistant-reportedly-pivoting-to-generative-ai/
One Image to High-Quality 3D Object Generation Magic123 is a PyTorch implementation that enables high-quality 3D object generation from a single image using both 2D and 3D diffusion priors. https://github.com/guochengqian/Magic123
A chat language model that can interpret and execute functions or plugins. It's designed to determine when to execute a function and understand its output, triggering functions only as needed. Function definitions are given as JSON Schema Objects, similar to OpenAI GPT function calls, and it's based on Llama 2. https://github.com/MeetKai/functionary
GetZep is avector database provider / open source project.
GitHub repo of Color-diffusion. It is a Python project that aims to colorize black and white images with SD
A blog post on Stack Overflow announcing the launch of OverflowAI, a new tool that uses machine learning to provide developers with instant answers to their coding questions.
GitHub repo of LP-Music-Caps. It uses a capsule network for music transcription.
Provides a set of tools for llm models. ToolBench official website
LLM-Attacks focuses on universal and transferable adversarial attacks on aligned language models.
GitHub repo of "AudioCraft". It is a library for audio generation
This is a project hosted on Hugging Face Spaces by Facebook. It is a project related to music generation
This is a GitHub repository for a project called "Jupyter AI". It is a generative AI extension for JupyterLab.