Upload 3 files

dpo/config.json

@@ -0,0 +1,12 @@
+{
+  "_class_name": "Showo",
+  "_diffusers_version": "0.30.1",
+  "codebook_size": 8192,
+  "llm_model_path": "microsoft/phi-1_5",
+  "llm_vocab_size": 50295,
+  "load_from_showo": false,
+  "mask_token_id": 58497,
+  "num_vq_tokens": 256,
+  "vocab_size": 58498,
+  "w_clip_vit": false
+}

dpo/config.yaml

@@ -0,0 +1,80 @@
+wandb:
+  entity: null
+  resume: auto
+  run_id: vkmnwd3v
+experiment:
+  project: training
+  name: show-o-dpo(only sft)
+  output_dir: /ssd7.7tb2/zrr/tcz/Show-o/ckts/train_set_hq_prune_full_t2i_dpo_10k_1e-5
+  save_every: 10000
+  eval_every: 2500
+  generate_every: 1000
+  log_every: 1
+  log_grad_norm_every: 20
+  logging_dir: /ssd7.7tb2/zrr/tcz/Show-o/ckts/train_set_hq_prune_full_t2i_dpo_10k_1e-5/logs
+model:
+  vq_model:
+    type: magvitv2
+    vq_model_name: showlab/magvitv2
+    pretrained_model_path: /ssd7.7tb2/zrr/tcz/Show-o/ckts/magvitv2/pytorch_model.safetensors
+  showo:
+    load_from_showo: false
+    pretrained_model_path: /ssd7.7tb2/zrr/tcz/Show-o/ckts/pytorch_model.safetensors
+    w_clip_vit: false
+    vocab_size: 58498
+    llm_vocab_size: 50295
+    llm_model_path: microsoft/phi-1_5
+    codebook_size: 8192
+    num_vq_tokens: 256
+    num_new_special_tokens: 10
+  gradient_checkpointing: true
+dataset:
+  gen_type: t2i
+  params:
+    data_path: /ssd7.7tb2/zrr/tcz/Show-o/ov/t2i_dpo_draft.yaml
+    validation_prompts_file: validation_prompts/showoprompts.txt
+    shuffle_buffer_size: 1000
+    num_workers: 32
+    resolution: 256
+    pin_memory: true
+    persistent_workers: true
+  preprocessing:
+    max_seq_length: 128
+    resolution: 256
+    center_crop: false
+    random_flip: false
+optimizer:
+  name: adamw
+  params:
+    learning_rate: 1.0e-05
+    scale_lr: false
+    beta1: 0.9
+    beta2: 0.999
+    weight_decay: 0.01
+    epsilon: 1.0e-08
+lr_scheduler:
+  scheduler: cosine
+  params:
+    learning_rate: ${optimizer.params.learning_rate}
+    warmup_ratio: 0.1
+training:
+  gradient_accumulation_steps: 1
+  batch_size_t2i: 10
+  mixed_precision: bf16
+  enable_tf32: true
+  seed: 10086
+  overfit_one_batch: false
+  cond_dropout_prob: 0.1
+  min_masking_rate: 0.0
+  label_smoothing: 0.0
+  max_grad_norm: null
+  guidance_scale: 0.0
+  generation_timesteps: 12
+  beta: 0.1
+  reward_coef: 0
+  dpo_coef: 1
+  sft_coef: 0
+  num_epoch: 1
+inference:
+  generation_timesteps: 18
+  guidance_scale: 1.75

dpo/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11b6ca792b5e4d0569b45a6f9511f422c3654d4276f5c1cb8ca2464eab4851cf
+size 2896413564