Search results

Inference time-step selection

Status: Open.

#21 In NVlabs/rcm;

· Turlan opened on Jan 22, 2026

Why don't use the F_pred of the student_F_withT() as F_theta_sg in loss?

Status: Open.

#19 In NVlabs/rcm;

· meng-ustc opened on Jan 21, 2026

Warmup ratio

Status: Open.

#18 In NVlabs/rcm;

· Turlan opened on Jan 20, 2026

Mad dmd loss

Status: Open.

#14 In NVlabs/rcm;

· phyllispeng123 opened on Jan 5, 2026

The denoise function output X0, XT ? why using XT as F_theta to calculate grad

Status: Open.

#13 In NVlabs/rcm;

· phyllispeng123 opened on Dec 31, 2025

how much memory are needed at least for training a 14B-parameter model?

Status: Open.

#12 In NVlabs/rcm;

· hsjkdjj opened on Dec 25, 2025

Discrepancy Between Scaling Factors in Paper and Code, and Typical Loss Magnitudes

Status: Open.

#8 In NVlabs/rcm;

· Peterande opened on Nov 4, 2025

rCM Huggingface demo

Status: Open.

#6 In NVlabs/rcm;

· wangyuji0606 opened on Oct 17, 2025

Is rcm support wan I2V model ?

Status: Open.

#2 In NVlabs/rcm;

· armynew opened on Oct 11, 2025