Free-Bloom
Free-Bloom copied to clipboard
[NeurIPS 2023] Free-Bloom: Zero-Shot Text-to-Video Generator with LLM Director and LDM Animator
Hi, I just followed your README and ran the flower example, but I got very noisy samples. Do you maybe know the reasons? 
This is the stacktrace of the issue: Num processes: 1 Process index: 0 Local process index: 0 Device: cuda Mixed precision type: fp16 {'resnet_time_scale_shift', 'dual_cross_attention', 'upcast_attention', 'class_embed_type', 'num_class_embeds', 'only_cross_attention', 'use_linear_projection'}...