UniSimModel<T>
Models & Types · AiDotNet.Diffusion.Video.WorldModels
UniSim universal simulator from video and action pairs.
Inheritance: DiffusionModelBase<T> → LatentDiffusionModelBase<T> → VideoDiffusionModelBase<T> → UniSimModel<T>
For Beginners
UniSim is a universal simulator that learns to simulate any environment from video and action pairs. It generalizes across different domains (robotics, games, real-world) to predict future observations.
How It Works
References:
- Paper: “UniSim: Learning Interactive Real-World Simulators” (2024)
UniSim is a universal simulator that learns from diverse real-world and simulated videos. Given a previous video segment and an action prompt, it predicts the continuation video through supervised learning. The model enables simulation of diverse real-world interactions for robotics, gaming, and embodied AI applications.
Technical specifications:
- Architecture: Universal Simulation DiT + Action-Video Pairs
- Latent channels: 16
- Default: 30 frames at 15 FPS
- Supports I2V: Yes | T2V: Yes | V2V: No
Example
var options = new DiffusionModelOptions<float> { LatentChannels = 16, DefaultInferenceSteps = 20 };
var actionInput = Tensor<float>.CreateRandom(new[] { 1, 16, 30, 32, 32 });
var trainX = Tensor<float>.CreateRandom(4, 8);
var trainY = Tensor<float>.CreateRandom(4, 2);
var result = new AiModelBuilder<float, Tensor<float>, Tensor<float>>()
.ConfigureModel(new UniSimModel<float>(options: options))
.Build(trainX, trainY);
var simulated = result.Predict(actionInput);
Constructors
| Constructor | Summary |
|---|---|
UniSimModel(NeuralNetworkArchitecture<T> architecture, DiffusionModelOptions<T> options, INoiseScheduler<T> scheduler, DiTNoisePredictor<T> predictor, TemporalVAE<T> temporalVAE, IConditioningModule<T> conditioner, int defaultNumFrames, int defaultFPS, int? seed) |
Initializes a new instance of UniSimModel with full customization support. |
Properties
| Property | Summary |
|---|---|
IConditioningModule<T> Conditioner { get; } |
|
INoisePredictor<T> NoisePredictor { get; } |
|
IVAEModel<T> TemporalVAE { get; } |
|
IVAEModel<T> VAE { get; } |
|
bool SupportsImageToVideo { get; } |
|
bool SupportsTextToVideo { get; } |
|
bool SupportsVideoToVideo { get; } |
|
int LatentChannels { get; } |
Methods
| Method | Summary |
|---|---|
ModelMetadata<T> GetModelMetadata() |