everything-ai @ v4.2.0🤖

CHANGELOG

• 🎨 fal-img2img: New task that lets you use fal.ai ComfyUI API to generate images starting from your PNG and JPEG images: you just need an API key! You can also customize the generation working with prompts and seeds
• 😎 Small adjustments in task selection interface

everything-ai @ v4.1.0🤖

CHANGELOG

• 🤫 simply-chatting (new task): chat directly with OpenAI, Groq, Cohere and Anthropic models (without RAG); customize your assistant with system template instructions, maximum number of output tokens and temperature
• 🪢 simply-chatting and build-your-llm are now integrated with LangFuse as an optional feature: you will then be able to see your models performances real-time on your LangFuse account when using them on everything-ai!

everything-ai @ v4.0.0🤖

CHANGELOG

• 💬 The task build-your-llm was added: you can now create a customizable chat LLM to interact with your Qdrant database with the power of Anthropic, Groq, OpenAI and Cohere models, just providing an API key! You can also set the temperature and the max number of output tokens
• 🤠 Small fixes in llama.cpp-and-qdrant and retrieval-text-generation RAG pipeline

everything-ai @ v3.0.0🤖

CHANGELOG

• 🦙 Added support for llama.cpp as inference engine to chat with your pdfs, backed by Qdrant as vector database
• 😵 Small bugfixes in retrieval_text_generation.py

everything-ai @ v2.0.1🤖

CHANGELOG

• 🧬everything-ai-proteinfold interface now shows the 3D structure of a protein along with its molecular one, no more static backbone images
• 🏋️everything-ai-autotrain now supports direct file uploading, without the need to specify the path

everything-ai @ v2.0.0🤖

CHANGELOG

New functionalities that were added since last release:

• 🎙️🔊 Handle audio files or microphone recordings, classifying or transcribing them with almost every audio-classification and automatic-speech-recognition model on Hugging Face Hub.
• 📽️ Generate video from text prompts with almost every text-to-video model on HuggingFace Hub (original architecture by Vasiliy Katsyka)
• 🧬 Predict the 3D structure of proteins from their amino-acidic sequence, with EsmFold by AI at Meta
• 🏋️ Finetune HF models on several downstream tasks with AutoTrain local integration (AutoTrain is developed by Abhishek Thakur)
• 🗣️ Unleash powerful LLMs and exploit larger database collections for RAG with the integration of Hugging Face Spaces API and Supabase PostgreSQL databases.

everything-ai @ v1.1.0🤖

CHANGELOG:

• Added the possibility to perform reverse image search based on an image dataset uploaded by the user and on Qdrant vector databases🔥
• Added GPU support for almost every model and task🚀
• Added a new python package in the Dockerfile: datasets🥳

everything-ai @ v1.0.0🤖

CHANGELOG:

1. Changed name from everything-rag to everything-ai
2. User-friendly task, model and file selection interface (no longer CLI, but GUI with gradio)
3. Multi-container application with docker compose up (simpler)
4. New docker image on Docker Hub: astrabert/everything-ai
5. Supports new tasks: text summarization, stable diffusion, image to text, image classification
6. Added multilingual support for text generation and stable diffusion
7. Text generation has now two options: agnostic text generation (no retrieval or prior knowledge) and retrieval text generation (retrieval with knowledge base)
8. Switched from Langchain-supplied ChromaDB local storage to qdrant collections for retrieval knowledge bases
9. Faster retrieval text generation inference

everything-rag @ v0.1.1🤖

Docker image is now stable and supports all tasks both on Windows and on Linux. HF space was added, more troubleshooting cases and tags that give at-a-glance stats on the status of this repo.

everything-rag @ v0.1.0🤖

Added Dockerfile and GradIO interface and the installable docker image as a package under the name ghcr.io/astrabert/everything-rag:latest.