flowmatch_scheduler_create

Create a FlowMatch Euler Discrete Scheduler

Description

Creates a FlowMatch scheduler for use with flow-matching diffusion models like LTX-2. FlowMatch schedulers use Euler integration for sampling, which is simpler and often faster than DDIM-style schedulers.

Usage

flowmatch_scheduler_create(num_train_timesteps = 1000L, shift = 1,
                           use_dynamic_shifting = FALSE, base_shift = 0.5,
                           max_shift = 1.15, base_seq_len = 256L,
                           max_seq_len = 4096L, invert_sigmas = FALSE,
                           shift_terminal = NULL,
                           time_shift_type = c("exponential", "linear"))

Arguments

num_train_timesteps: Integer. The number of diffusion steps used to train the model. Default: 1000
shift: Numeric. The shift value for the timestep schedule. Default: 1.0
use_dynamic_shifting: Logical. Whether to apply timestep shifting on-the-fly based on the image/video resolution. Default: FALSE
base_shift: Numeric. Value to stabilize generation. Increasing reduces variation. Default: 0.5
max_shift: Numeric. Maximum shift allowed. Increasing encourages more variation. Default: 1.15
base_seq_len: Integer. Base sequence length for dynamic shifting. Default: 256
max_seq_len: Integer. Maximum sequence length for dynamic shifting. Default: 4096
invert_sigmas: Logical. Whether to invert the sigmas (used by some models like Mochi). Default: FALSE
shift_terminal: Numeric or NULL. End value of shifted schedule. Default: NULL
time_shift_type: Character. Type of dynamic shifting: “exponential” or “linear”. Default: “exponential”

Details

FlowMatch (Flow Matching) is a framework for training continuous normalizing flows by regressing onto target probability paths. The Euler discrete scheduler implements simple Euler integration for sampling from trained flow models.

The core update rule is: prev_sample = sample + dt * model_output where dt = sigma_next - sigma_current.

Value

A FlowMatch scheduler object (list) containing:

sigmas: The noise schedule
timesteps: The timestep schedule
num_train_timesteps: Training timesteps
config: All configuration parameters

References

Lipman, Y., Chen, R. T. Q., Ben-Hamu, H., Nickel, M., & Le, M. (2022). “Flow Matching for Generative Modeling.” https://arxiv.org/abs/2210.02747

Examples

# Create a FlowMatch scheduler
scheduler <- flowmatch_scheduler_create(
  num_train_timesteps = 1000,
  shift = 1.0
)

# Set timesteps for inference
scheduler <- flowmatch_set_timesteps(scheduler, num_inference_steps = 8)

Packages