TheMarco

Global seed set to 23
Running on GPUs 0,
Loading model from model.ckpt
LatentDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.52 M params.
making attention of type 'vanilla' with 512 in_channels
Working with z of shape (1, 4, 64, 64) = 16384 dimensions.
making attention of type 'vanilla' with 512 in_channels
Some weights of the model checkpoint at openai/clip-vit-large-patch14 were not used when initializing CLIPTextModel: ['vision_model.encoder.layers.8.self_attn.out_proj.bias', 'vision_model.encoder.layers.13.mlp.fc2.bias', 'vision_model.encoder.layers.4.self_attn.v_proj.weight', 'vision_model.encoder.layers.14.mlp.fc1.weight', 'vision_model.encoder.layers.9.layer_norm1.bias', 'vision_model.encoder.layers.16.self_attn.q_proj.weight', 'vision_model.encoder.layers.4.self_attn.q_proj.bias', 'vision_model.encoder.layers.23.mlp.fc1.bias', 'vision_model.pre_layrnorm.bias', 'vision_model.encoder.layers.8.mlp.fc2.bias', 'vision_model.encoder.layers.19.self_attn.k_proj.bias', 'vision_model.encoder.

TheMarco

Verifying my Blockstack ID is secured with the address 1Q1Wi8Nu4tjnM8VgCj7NHwwq4rBNgtmwTf https://explorer.blockstack.org/address/1Q1Wi8Nu4tjnM8VgCj7NHwwq4rBNgtmwTf

TheMarco

/*

urldecoder.js
This simple service reveals the real URL behind a shortened one

Usage: fire it up with:

node urldecoder.js

Then access as follows: