Models

Mistral AI provides frontier and open models plus a developer platform for generative AI apps.

Microsoft Phi is a family of small language models for efficient generative AI workloads.

TabPFN is a tabular foundation model package for fast supervised learning on small tabular datasets.

Chronos is Amazon Science's family of pretrained probabilistic models for time series forecasting.

Darts is a Python library for user friendly forecasting and anomaly detection on time series.

GluonTS is a Python toolkit for probabilistic time series modeling and forecasting.

Lag-Llama is a probabilistic foundation model for time series forecasting.

Moirai is a universal time series forecasting model distributed through Salesforce's Uni2TS project, including the sparse mixture-of-experts Moirai-MoE variant.

MOMENT is a family of open source foundation models for general purpose time series analysis.

N-BEATS is a neural architecture for interpretable and accurate time series forecasting.

N-HiTS is a neural hierarchical interpolation model for efficient long horizon time series forecasting.

PatchTST is a transformer based model for multivariate time series forecasting and representation learning.

Prophet is a forecasting procedure for time series data with trend seasonality and holiday effects.

sktime is a unified Python framework for machine learning with time series.

Temporal Fusion Transformer is a Google Research model for interpretable multi horizon time series forecasting.

Time-MoE is a scalable mixture of experts foundation model for time series forecasting.

TimeGPT is Nixtla's generative pretrained model for time series forecasting and analytics.

Timer-XL is part of THUML's large time series model project for foundation model forecasting.

TimesFM is a Google Research time series foundation model for forecasting.

TimesNet is a time series analysis model that captures temporal variations with 2D tensors.

Tiny Time Mixers are compact IBM Granite time series models for efficient forecasting.

Toto is Datadog's open source time series foundation model for forecasting observability signals.

Neural network structure for adding spatial conditioning controls to large diffusion models.

World foundation model platform for physical AI development using video and simulation data

NVIDIA research implementation of StyleGAN3 for alias free generative adversarial image synthesis.

World Labs model for generating controllable 3D worlds and interactive spatial scenes from prompts

LambdaMART is a gradient-boosted learning-to-rank approach implemented by tools such as LightGBM's LambdaRank objective.

monoT5 is a T5-based neural reranking model used through PyGaggle for passage and document ranking experiments.

Official implementation for real time radiance field rendering with 3D Gaussian splatting.

Google DeepMind model for predicting biomolecular structure and interactions across proteins DNA RNA and ligands

Open biomolecular foundation model for structure prediction and binding affinity modeling

Chai Discovery model for predicting molecular structures and biomolecular interactions for drug discovery

Generative protein language model for reasoning over protein sequence structure and function

Arc Institute genomic foundation model for long-context DNA sequence modeling and biological design.

DeepMind diffusion-based weather model for probabilistic medium-range forecasting and extreme-weather risk prediction.

Alphabet company building AI models for drug discovery and biomedical research.

Neural Radiance Fields research project for synthesizing novel views from sparse posed images.

Collaborative benchmark suite with many tasks for probing and evaluating language models.

GIFT-Eval is a Salesforce benchmark and leaderboard for evaluating general time series forecasting models.

Graduate level Google proof Q&A benchmark for evaluating hard scientific reasoning by language models.

Dataset of grade school math word problems for evaluating multi step reasoning in language models.

Commonsense natural language inference benchmark with adversarially filtered endings.

OpenAI code generation benchmark of Python programming problems for functional correctness evaluation.

Open platform for community based side by side evaluation and ranking of AI models.

Massive Multitask Language Understanding benchmark for measuring broad model knowledge.

More robust and challenging MMLU style benchmark dataset for advanced language model evaluation.