Valence Labs Courses

Spatiotemporal Modeling of Molecular Holograms

Explore Spateo, a 3D spatiotemporal modeling framework for embryogenesis, enabling molecular hologram creation, multi-level biological analysis, and insights into organ development through advanced computational techniques.

• YouTube
• 1 hour 9 minutes
• Self-Paced
• Free Video

Distilling Foundation Models via Energy Hessians for Fast, Specialized Machine Learning Force Fields

Explore the distillation of foundation models for faster, specialized Machine Learning Force Fields in computational chemistry, focusing on energy Hessians to maintain accuracy while improving speed.

• YouTube
• 54 minutes
• Self-Paced
• Free Video

Improving the Speed and Accuracy of Neural Network Interatomic Potentials Across Chemical Domains

Explore how scaling strategies can enhance Neural Network Interatomic Potentials (NNIPs), focusing on the Efficiently Scaled Attention Interatomic Potential (EScAIP) that achieves faster inference and better performance across chemical domains.

• YouTube
• 54 minutes
• Self-Paced
• Free Video

GENERator: A Long-Context Generative Genomic Foundation Model

Discover how GENERator, a generative genomic foundation model, decodes DNA sequences with a 98k base pair context length, demonstrating state-of-the-art performance in genomic research and biotechnological applications.

• YouTube
• 42 minutes
• Self-Paced
• Free Video

HELM: Hierarchical Encoding for mRNA Language Modeling

Dive into hierarchical encoding for mRNA language modeling, exploring how HELM's novel pre-training strategy incorporates codon-level structure to improve biological sequence analysis and outperform standard models.

• YouTube
• 59 minutes
• Self-Paced
• Free Video

Foundation Models for Biological Data Modalities - Advances in Genomic and Protein Sequence Analysis

Explore cutting-edge foundation models in biological sequence analysis, covering genomic language models, multimodal conversational agents, and transfer learning between DNA, RNA, and protein data.

• YouTube
• 1 hour 5 minutes
• Self-Paced
• Free Video

Does Equivariance Matter at Scale? A Study of Neural Architectures and Symmetries

Explore the scaling dynamics of equivariant and non-equivariant neural networks in rigid-body interactions, examining their performance, data efficiency, and computational trade-offs.

• YouTube
• 1 hour 4 minutes
• Self-Paced
• Free Video

The Superposition of Diffusion Models Using the Itô Density Estimator

Explore groundbreaking techniques for combining pre-trained diffusion models using the Itô density estimator, enabling efficient image generation and protein structure design without retraining.

• YouTube
• 1 hour 1 minute
• Self-Paced
• Free Video

Targeting Protein-Ligand Neosurfaces with a Generalizable Deep Learning Tool

Explore groundbreaking deep learning approaches for designing proteins that target drug-bound protein complexes, with applications in drug-controlled cell therapies and synthetic biology.

• YouTube
• 1 hour 11 minutes
• Self-Paced
• Free Video

Beyond Sequence - Impact of Geometric Context for RNA Property Prediction

Explore how geometric context enhances RNA property prediction through advanced deep learning models, improving accuracy and efficiency in drug discovery and biological research.

• YouTube
• 54 minutes
• Self-Paced
• Free Video

Single-cell Multiomics Data Integration and Generation with scPairing

Explore single-cell multiomics data integration with scPairing, a variational autoencoder model that embeds different cellular modalities into a common space, enabling generation of novel multiomics data without costly technologies.

• YouTube
• 56 minutes
• Self-Paced
• Free Video

Orb-v3: Atomistic Simulation at Scale

Explore Orb-v3, a next-generation universal interatomic potential that expands the performance-speed-memory frontier with 10x reduced latency and 8x reduced memory while maintaining near state-of-the-art accuracy for atomic simulations.

• YouTube
• 1 hour 14 minutes
• Self-Paced
• Free Video

FEAT: Free Energy Estimators with Adaptive Transport

Explore free energy estimation through the FEAT framework, which uses adaptive transport to provide consistent estimators and variational bounds, unifying equilibrium and non-equilibrium methods for neural free energy calculations.

• YouTube
• 1 hour 1 minute
• Self-Paced
• Free Video

Mapping Cells Through Time and Space with Moscot

Explore the moscot framework for single-cell genomics, enabling multimodal analysis across temporal and spatial dimensions to reconstruct developmental trajectories and uncover spatiotemporal dynamics in biological systems.

• YouTube
• 1 hour 8 minutes
• Self-Paced
• Free Video

A General Framework for Inference-time Scaling and Steering of Diffusion Models

Explore a general framework for inference-time scaling and steering of diffusion models, enabling user-specified properties without expensive fine-tuning through the Feynman Kac steering method.

• YouTube
• 1 hour 18 minutes
• Self-Paced
• Free Video