Skip to content

What to put in pretrained model path ? #2

Description

@Quantum-MIT

pretrained_model_path: "cerspense/zeroscope_v2_576w"
On setting this getting the key issue:

ValueError: Cannot load <class 'lamp.models.unet.UNet3DConditionModel'> from cerspense/zeroscope_v2_576w because the following keys are missing:
up_blocks.2.attentions.1.transformer_blocks.0.attn_temp.to_out.0.weight, down_blocks.2.downsamplers.0.conv.conv1d.weight, mid_block.resnets.0.conv1.conv2d.bias, mid_block.resnets.1.conv2.conv2d.bias, up_blocks.3.resnets.1.conv_shortcut.conv_gate.bias, up_blocks.2.resnets.1.conv_shortcut.conv2d.bias, down_blocks.0.resnets.1.conv1.conv_gate.bias, up_blocks.2.resnets.1.conv_shortcut.conv1d.bias, down_blocks.1.resnets.1.conv2.conv_gate.bias, up_blocks.0.resnets.1.conv_shortcut.conv_gate.bias, down_blocks.2.attentions.1.transformer_blocks.0.norm_temp.weight, up_blocks.3.resnets.1.conv1.conv2d.weight, up_blocks.3.attentions.1.transformer_blocks.0.norm_temp.bias, up_blocks.2.attentions.1.transformer_blocks.0.attn_temp.to_q.weight, up_blocks.3.attentions.1.transformer_blocks.0.attn_temp.to_q.weight, up_blocks.3.resnets.0.conv2.conv2....

Also, directly running python moving_test.py --model zeroscope_v2_576w --checkpoint_folder ./outputs/train/riding_bicycle --checkpoint_index 300 --noise_prior 0. --seed 7192280 --sdp --prompt "A cat is running on the grass." --token_indices 2 --bbox [[0,208,192,376]] --attnres 12 --repeat_num 3 --boxdiff 1 --moving --temp --attnloss 0.1 --posloss 0.05

doesnt work as ./outputs/train/ folder isn't created yet. So we need to run moving_train.py first ?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions