tokenizer/vqgan.yaml

model:
  base_learning_rate: 4.5e-06
  target: taming.models.vqgan.VQModel
  params:
    embed_dim: 256
    n_embed: 8192
    ddconfig:
      double_z: false
      z_channels: 256
      resolution: 512
      in_channels: 3
      out_ch: 3
      ch: 128
      ch_mult:
      - 1
      - 1
      - 2
      - 2
      - 4
      num_res_blocks: 2
      attn_resolutions: []
      dropout: 0.0
    lossconfig:
      target: taming.modules.losses.vqperceptual_vit_vqgan.VQLPIPSWithDiscriminator
      params:
        disc_start: 100001
        perceptual_weight: 1.0
        adversarial_weight: 0.5
        disc_params:
          size: 512
    ckpt_path: manifold://fair_onellm_checkpoints/tree/v2/tokenizer/vqgan_wm_0209.ckpt
data:
  target: main.DataModuleFromConfig
  params:
    batch_size: 4
    num_workers: 10
    image_size: 512
    filter_image_size: 512
    dataset: coco
    aesthetics_th: 0
    clipsim_th: 0
--distributed-world-size: null
'32': null
--distributed-port: null
'17338': null
--save-dir: null
/checkpoint/shellysheynin/shutterstock/512x512_1024tokens_4node_shutterstock_laion_no_attn_styleGAN:
  log_every-500:
    ngpu32: null
--tensorboard-logdir: null
/checkpoint/shellysheynin/tensorboard_logs/2023-03-30/512x512_1024tokens_4node_shutterstock_laion_no_attn_styleGAN:
  log_every-500:
    ngpu32: null
'14561': null
/checkpoint/shellysheynin/tensorboard_logs/2023-04-02/512x512_1024tokens_4node_shutterstock_laion_no_attn_styleGAN:
  log_every-500:
    ngpu32: null