Translations:Diffusion Models Are Real-Time Game Engines/68/zh: Difference between revisions

Latest revision as of 00:28, 9 September 2024

Information about message (contribute)

This message has no documentation. If you know where or how this message is used, you can help other translators by adding documentation to this message.

Message definition (Diffusion Models Are Real-Time Game Engines)

To ablate the impact of noise augmentation, we train a model without added noise. We evaluate both our standard model with noise augmentation and the model without added noise (after 200k training steps) auto-regressively and compute PSNR and LPIPS metrics between the predicted frames and the ground-truth over a random holdout of 512 trajectories. We report average metric values for each auto-regressive step up to a total of 64 frames in Figure [https://arxiv.org/html/2408.14837v1#S5.F7 7].

为了消除噪声增强的影响，我们训练了一个不添加噪声的模型。我们对标准噪声增强模型和不添加噪声的模型（经过 200,000 步训练后）进行自回归评估，并计算在随机保留的 512 条轨迹上预测帧与真实帧之间的 PSNR 和 LPIPS 指标。我们在图 7 中报告了每个自回归步骤的平均指标值，最多可达 64 帧。