Pretrained dalle.pt model file

[tianye2856]

Hello, Thanks for your amazing work. I am a Mac user and does not have GPU available to train a new model. Could you please provide the pretrained dalle.pt model. So I can generate some images from text, and test it. Thanks in advance.

[johnpaulbin]

Hey there @tianye2856 , us at the Dall-E Discord server has some models released! See here (colab included, so you don't need to use your own computer to inference): https://github.com/robvanvolt/DALLE-models

[batmanscode]

Hey there @tianye2856 , us at the Dall-E Discord server has some models released! See here (colab included, so you don't need to use your own computer to inference): https://github.com/robvanvolt/DALLE-models

This is awesome! Maybe it should be included in the readme so it's easy to find 😃

[johnpaulbin]

Hey there @tianye2856 , us at the Dall-E Discord server has some models released! See here (colab included, so you don't need to use your own computer to inference): https://github.com/robvanvolt/DALLE-models

This is awesome! Maybe it should be included in the readme so it's easy to find 😃

Now it has!

[kartikpodugu]

I am trying to use the models at the following link. https://github.com/robvanvolt/DALLE-models/tree/main/models/taming_transformer/8L_64HD_8H_756I_128T_cc12mcc3m_2E I am getting the following error with latest repository as on 17th May 2023.

From what I understand, model when pt file is generated is different from now. How to find the model for the corresponding pt file ?? Infact, I am getting similar errors for all pre-trained models available. Anybody has any other pre trained pt models with current respository?

size mismatch for image_emb.weight: copying a param with shape torch.Size([1024, 512]) from checkpoint, the shape in current model is torch.Size([8192, 512]).
size mismatch for image_pos_emb.weights_0: copying a param with shape torch.Size([1, 16, 1, 512]) from checkpoint, the shape in current model is torch.Size([1, 32, 1, 512]).
size mismatch for image_pos_emb.weights_1: copying a param with shape torch.Size([1, 1, 16, 512]) from checkpoint, the shape in current model is torch.Size([1, 1, 32, 512]).
size mismatch for transformer.layers.blocks.0.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.1.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.2.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.3.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.4.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.5.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.6.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.7.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.8.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.9.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.10.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.11.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.12.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.13.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.14.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for transformer.layers.blocks.15.f.net.fn.fn.attn_fn.master_layout: copying a param with shape torch.Size([8, 24, 24]) from checkpoint, the shape in current model is torch.Size([8, 72, 72]).
size mismatch for to_logits.1.weight: copying a param with shape torch.Size([50560, 512]) from checkpoint, the shape in current model is torch.Size([57728, 512]).
size mismatch for to_logits.1.bias: copying a param with shape torch.Size([50560]) from checkpoint, the shape in current model is torch.Size([57728]).