Foundation Models
Research papers, repositories, and articles about foundation models
Showing 8 of 8 items
google-research/timesfm
Time-series foundation model from Google Research. Lets you forecast many different signals with one shared model, not one model per metric.
HeartMuLa: A Family of Open Sourced Music Foundation Models
HeartMuLa bundles an audio–text matcher, robust lyric recognizer, music codec, and a music-generating LLM. You get controllable, prompt-driven song generation plus tools for indexing and understanding songs at scale.
Robbyant/lingbot-map
Implements a 3D "foundation model" that reconstructs scenes from streaming sensor data in a simple feed-forward pass. Good reference if you're exploring continuous 3D perception for agents.
Next-Embedding Prediction Makes Strong Vision Learners
Instead of predicting pixels or patches, this method predicts the next embedding in a learned space. Vision folks can plug this into pretraining to squeeze more out of ImageNet-scale data.
Depth Any Panoramas: A Foundation Model for Panoramic Depth Estimation
Depth Any Panoramas builds a single model for depth on 360° indoor and outdoor scenes. Robotics and AR teams can reuse this instead of training per-dataset depth nets.
Unified Zero-Shot Time Series Forecasting: A Darts Foundation
Plugs several time-series foundation models into the Darts library under one standard interface. Lets teams swap in Chronos-2, TimesFM 2.5, TiRex, and PatchTST-FM with a name change. If you forecast anything, you can now A/B strong zero-shot models without glue code. ([arxiv.org](https://arxiv.org/list/cs.LG/new))
PairSAE: Mechanistic Interpretability from Pair Representations in Protein Co-Folding
Adapts sparse autoencoders to the "pair" tensors in protein co-folding models by compressing them into token-level features first. Recovers features aligned with biological structure and binding signals. If you care about interpretability beyond plain transformers, this is a useful template. ([arxiv.org](https://arxiv.org/list/cs.LG/new))
Do Foundation Models Know Geometry? Probing Frozen Features for Continuous Physical Measurement
Probes frozen vision backbones for tasks like measuring lengths and angles. Tests how much real-world geometry is already baked into general-purpose models.