sample seq problem

[redcping]

In the diffusion.py file, during uniform sampling, the sequence is obtained as follows:
skip = self.num_timesteps // self.args.timesteps
seq = range(0, self.num_timesteps, skip)
However, when sampling 10 steps, the generated sequence is [0, 100, 200, 300, 400, 500, 600, 700, 800, 900]. This means the denoising starts from t=900. Shouldn't the sequence be [0, 111, 222, 333, 444, 555, 666, 777, 888, 999] to ensure a more evenly spaced sampling?

[qiaozhijian]

See this issue 686. It seems starting from 1000 is better.