Making large AI models cheaper, faster and more accessible

[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

Janus-Series: Unified Multimodal Understanding and Generation Models

YuE: Open Full-song Music Generation Foundation Model, something similar to Suno.ai but open

Data processing for and with foundation models! 🍎 🍋 🌽 ➡️ ➡️🍸 🍹 🍷

⚡ TabPFN: Foundation Model for Tabular Data ⚡

Chronos: Pretrained Models for Time Series Forecasting

DeepSeek-VL: Towards Real-World Vision-Language Understanding

Code and models for ICML 2024 paper, NExT-GPT: Any-to-Any Multimodal Large Language Model

TimeGPT-1: production ready pre-trained Time Series Foundation Model for forecasting and anomaly detection. Generative pretrained transformer for time series trained on over 100B data points. It's capable of accurately predicting various domains such as retail, electricity, finance, and IoT with just a few lines of code 🚀.

[CVPR2024 Highlight][VideoChatGPT] ChatGPT with video understanding! And many more supported LMs such as miniGPT4, StableLM, and MOSS.

🦦 Otter, a multi-modal model based on OpenFlamingo (open-sourced version of DeepMind's Flamingo), trained on MIMIC-IT and showcasing improved instruction-following and in-context learning ability.

SuperCLUE: 中文通用大模型综合性基准 | A Benchmark for Foundation Models in Chinese

EVA Series: Visual Representation Fantasies from BAAI

Images to inference with no labeling (use foundation models to train supervised models).

Awesome things about LLM-powered agents. Papers / Repos / Blogs / ...

Prompt Learning for Vision-Language Models (IJCV'22, CVPR'22)

[ECCV2024] Video Foundation Models & Data for Multimodal Understanding

An automatic evaluator for instruction-following language models. Human-validated, high-quality, cheap, and fast.