same prompt+seed and everything. why the results are different between diffuser's and sd-webui?

[aiXia121]

Using the same seed, prompt and other text to generate image information, why are the images generated by the diffuser library different from the images generated by the stable diffuison webui, and the differences are quite big. Moreover, the effect of stable diffusion webui has always been better than that of diffuser for reasoning. What can be done to make the quality of images generated by diffuser for inference as good as that of stable diffusion webui?

thx