Upload viz/preference/pref_predict.yaml with huggingface_hub

viz/preference/pref_predict.yaml

@@ -0,0 +1,135 @@
+# lightning.pytorch==2.4.0
+seed_everything: true
+tags:
+  exp: &exp sftpix512ema
+torch_hub_dir: ~/.cache/torch/hub
+huggingface_cache_dir: null
+trainer:
+  default_root_dir: ./workdir_dpg_base
+  accelerator: auto
+  strategy: ddp
+  devices: auto
+  num_nodes: 1
+  accumulate_grad_batches: 4
+  precision: bf16-mixed
+  logger:
+      class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: ./universal_pix_flow_t2i
+        name: *exp
+  num_sanity_val_steps: 2
+  max_steps: 340000
+  gradient_clip_val: 1.0
+  gradient_clip_algorithm: norm
+  val_check_interval: 40000
+  check_val_every_n_epoch: null
+  log_every_n_steps: 50
+  deterministic: null
+  inference_mode: true
+  use_distributed_sampler: false
+  callbacks:
+    - class_path: src.callbacks.model_checkpoint.CheckpointHook
+      init_args:
+        every_n_train_steps: 10000
+        save_top_k: -1
+        save_last: true
+    - class_path: src.callbacks.save_images.SaveImagesHook
+      init_args:
+         save_dir: val
+  plugins:
+    - src.plugins.bd_env.BDEnvironment
+model:
+  vae:
+    class_path: src.models.autoencoder.pixel.PixelAE
+  denoiser:
+    class_path: src.models.transformer.dit_t2i_DeCo.PixNerDiT
+    init_args:
+      in_channels: 3
+      patch_size: 16
+      num_groups: 24
+      hidden_size: &hidden_dim 1536
+      txt_embed_dim: &txt_embed_dim 2048
+      txt_max_length: 128
+      num_text_blocks: 4
+      decoder_hidden_size: 32
+      num_encoder_blocks: 16
+      num_decoder_blocks: 3
+  conditioner:
+    class_path: src.models.conditioner.qwen3_text_encoder.Qwen3TextEncoder
+    init_args:
+      weight_path: Qwen/Qwen3-1.7B
+      embed_dim: *txt_embed_dim
+      max_length: 128
+  diffusion_trainer:
+    class_path: src.diffusion.flow_matching.training_repa_DeCo.REPATrainer
+    init_args:
+      lognorm_t: true
+      feat_loss_weight: 0.5
+      timeshift: 4.0
+      encoder:
+        class_path: src.models.encoder.DINOv2
+        init_args:
+          weight_path: /root/.cache/torch/hub/checkpoints/dinov2_vitb14_pretrain.pth
+      align_layer: 6
+      proj_denoiser_dim: *hidden_dim
+      proj_hidden_dim: *hidden_dim
+      proj_encoder_dim: 768
+      scheduler: &scheduler src.diffusion.flow_matching.scheduling.LinearScheduler
+  diffusion_sampler:
+    class_path: src.diffusion.flow_matching.adam_sampling.AdamLMSampler
+    init_args:
+      num_steps: 25
+      guidance: 4.0
+      timeshift: 3.0 # 3.0 for geneval
+      order: 2
+      scheduler: *scheduler
+      guidance_fn: src.diffusion.base.guidance.simple_guidance_fn
+  ema_tracker:
+    class_path: src.callbacks.simple_ema.SimpleEMA
+    init_args:
+      decay: 0.9999
+  optimizer:
+    class_path: torch.optim.AdamW
+    init_args:
+      lr: 1e-5
+      betas:
+        - 0.9
+        - 0.999
+      weight_decay: 0.00
+data:
+  train_dataset:
+    class_path: src.data.dataset.image_txt.ImageText
+    init_args:
+      root: /data/datasets/BLIP-3o/BLIP3o-60k
+      resolution: 512
+  # train_dataset:
+  #   class_path: src.data.dataset.blip3o_dataset.WebDatasetPackedDataset
+  #   init_args:
+  #     urls: [/data/datasets/BLIP-3o/BLIP3o-60k] # [/data/datasets/BLIP-3o/BLIP3o-Pretrain-JourneyDB,]
+  #     resolution: 512
+  #     random_crop: false
+  #     shuffle_buffer: 2000
+  #     sample_shuffle: true
+  #     repeat: true
+  eval_dataset:
+    class_path: src.data.dataset.geneval.GenEvalDataset
+    init_args:
+      meta_json_path: ./evaluations/geneval/evaluation_metadata.jsonl
+      num_samples_per_instance: 1
+      latent_shape:
+        - 3
+        - 512
+        - 512
+  pred_dataset:
+    class_path: src.data.dataset.dpg.DPGDataset
+    init_args:
+      prompt_path: /mnt/localssd/colligo/contrib/Mori/collections/foundation_exploration/configs/genrender6/ganloss/DeCo/evaluations/dpg/prompts
+      num_samples_per_instance: 1
+      latent_shape:
+        - 3
+        - 512
+        - 512
+  train_batch_size: 16
+  train_num_workers: 8
+  pred_batch_size: 8
+  pred_num_workers: 1