Upload 2 files

cifar-gqgan-gs64-cb1024/checkpoints/last.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70945089febc5f99c8ce821da6f0b7e0e994fe55e73e993cadc282ea925cd4ca
+size 498383136

cifar-gqgan-gs64-cb1024/config.yaml

@@ -0,0 +1,270 @@
+# lightning.pytorch==2.2.0
+seed_everything: 0
+trainer:
+  accelerator: gpu
+  strategy: ddp_find_unused_parameters_true
+  devices: 6
+  num_nodes: 1
+  precision: 16-mixed
+  logger:
+    class_path: lightning.pytorch.loggers.TensorBoardLogger
+    init_args:
+      save_dir: ./logs/gqgan/
+      name: null
+      version: cifar-test1
+      log_graph: false
+      default_hp_metric: true
+      prefix: ''
+      sub_dir: null
+      comment: ''
+      purge_step: null
+      max_queue: 10
+      flush_secs: 120
+      filename_suffix: ''
+  callbacks:
+  - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+    init_args:
+      dirpath: ./logs/gqgan/cifar-test1/checkpoints/
+      filename: null
+      monitor: null
+      verbose: false
+      save_last: null
+      save_top_k: 1
+      save_weights_only: false
+      mode: min
+      auto_insert_metric_name: true
+      every_n_train_steps: null
+      train_time_interval: null
+      every_n_epochs: null
+      save_on_train_epoch_end: null
+      enable_version_counter: true
+  - class_path: lightning.pytorch.callbacks.LearningRateMonitor
+    init_args:
+      logging_interval: step
+      log_momentum: false
+      log_weight_decay: false
+  fast_dev_run: false
+  max_epochs: 30
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  max_time: null
+  limit_train_batches: null
+  limit_val_batches: null
+  limit_test_batches: null
+  limit_predict_batches: null
+  overfit_batches: 0.0
+  val_check_interval: null
+  check_val_every_n_epoch: 1
+  num_sanity_val_steps: -1
+  log_every_n_steps: 100
+  enable_checkpointing: null
+  enable_progress_bar: null
+  enable_model_summary: null
+  accumulate_grad_batches: 1
+  gradient_clip_val: null
+  gradient_clip_algorithm: null
+  deterministic: null
+  benchmark: null
+  inference_mode: true
+  use_distributed_sampler: true
+  profiler: null
+  detect_anomaly: false
+  barebones: false
+  plugins: null
+  sync_batchnorm: false
+  reload_dataloaders_every_n_epochs: 0
+  default_root_dir: null
+model:
+  class_path: gstk.models.gqgan_vq_v1.GQGAN
+  init_args:
+    fm_shape:
+    - 16
+    - 16
+    img_encoder_cfg:
+      double_z: false
+      z_channels: 64
+      resolution: 128
+      in_channels: 3
+      out_ch: 3
+      ch: 128
+      ch_mult:
+      - 1
+      - 2
+      num_res_blocks: 4
+    img_decoder_cfg:
+      double_z: false
+      z_channels: 3
+      resolution: 128
+      in_channels: 3
+      out_ch: 3
+      ch: 128
+      ch_mult:
+      - 1
+      - 2
+      num_res_blocks: 4
+    z_channels: 3
+    vq_cfg:
+      learnable_codebook: false
+      ema_update: true
+      sync_codebook: true
+      dim: 3
+      codebook_size: 1024
+      accept_image_fmap: false
+      decay: 0.8
+      commitment_weight: 1.0
+      threshold_ema_dead_code: 2
+      kmeans_init: true
+      kmeans_iters: 10
+    gs_embed_cfg:
+      fm_lifter_cfg:
+        in_dim: 64
+        pe_version: sine
+        out_dim: 64
+      gaussian_lifter_cfg:
+        num_anchor: 64
+        embed_dim: 64
+        anchor_grad: true
+        ins_feat_grad: false
+        feature_dim: 64
+        include_opa: false
+      anchor_encoder_cfg:
+        embed_dim: 64
+        include_opa: false
+        feature_dim: 64
+      attn_encoder_cfg:
+        fm_shape:
+        - 16
+        - 16
+        encoder_layer_cfg:
+          embed_dim: 64
+          dropout: 0.1
+          activation: relu
+          n_levels: 1
+          num_heads: 4
+          n_points: 4
+        num_layers: 2
+      attn_decoder_cfg:
+        fm_shape:
+        - 16
+        - 16
+        decoder_layer_cfg:
+          embed_dim: 64
+          dropout: 0.1
+          activation: relu
+          n_levels: 1
+          num_heads: 4
+          n_points: 4
+        num_layers: 2
+        proj_drop: 0.1
+        residual_mode: cat
+      ffn_cfg:
+        in_channels: 128
+        pre_norm: true
+        out_norm: false
+        embed_dim: 64
+        num_fcs: 2
+        ffn_drop: 0.0
+        add_identity: true
+      refine_cfg:
+        embed_dim: 64
+        xy_range:
+        - -1
+        - 1
+        initial_scale_range:
+        - 0.1
+        - 4.0
+        final_scale_range:
+        - 0.1
+        - 3.0
+        refine_state:
+        - 0
+        - 1
+        include_opa: false
+        dim_feature: 64
+        z_channels: 3
+      spconv_cfg:
+        in_channels: 64
+        out_channels: 64
+        kernel_size: 5
+        fm_shape:
+        - 16
+        - 16
+        use_out_proj: false
+      operation_order:
+      - cross_attn
+      - ffn
+      - refine
+      - cross_attn
+      - ffn
+      - refine
+      - cross_attn
+      - ffn
+      - refine
+      op_param_share: false
+    image_key: image
+    use_ema: true
+    loss_cfg:
+      disc_conditional: false
+      disc_in_channels: 3
+      disc_start_epoch: 0
+      gan_start_epoch: 5
+      disc_weight: 0.8
+      gen_loss_weight: 0.1
+      lecam_loss_weight: 0.05
+      commit_weight: 0.25
+    optim_gen_cfg:
+      opt: adamw
+      lr: 0.0002
+      betas:
+      - 0.9
+      - 0.999
+      weight_decay: 0.01
+    optim_disc_cfg:
+      opt: adamw
+      lr: 0.0002
+      betas:
+      - 0.9
+      - 0.999
+      weight_decay: 0.01
+    lr_scheduler_cfg:
+      sched: cosine
+      warmup_lr: 1.0e-06
+      min_lr: 0.0
+      warmup_epochs: 1
+      step_on_epochs: false
+      num_epochs: null
+      updates_per_epoch: null
+    embed_dim: 64
+    num_gs: 64
+    xy_range:
+    - -1
+    - 1
+    initial_scale_range:
+    - 0.1
+    - 4.0
+    final_scale_range:
+    - 0.1
+    - 3.0
+    include_opa: false
+data:
+  class_path: gstk.data.from_config.DataModuleFromConfig
+  init_args:
+    batch_size: 64
+    train:
+      target: gstk.data.dataset.MyCIFAR
+      params:
+        train: true
+    validation:
+      target: gstk.data.dataset.MyCIFAR
+      params:
+        train: false
+    test:
+      target: gstk.data.dataset.MyCIFAR
+      params:
+        train: false
+    wrap: false
+    num_workers: 8
+optimizer: null
+lr_scheduler: null
+ckpt_path: null