Upload robocasa_step100000/config.yaml with huggingface_hub

robocasa_step100000/config.yaml

@@ -0,0 +1,47 @@
+datasets:
+  vla_data:
+    data_mix: robocasa_v02_all
+    data_root_dir: /home/sxu1/data/datasets/robocasa_v01_cosmos_lerobot
+    dataset_py: lerobot_datasets
+    delete_pause_frame: false
+    per_device_batch_size: 8
+    sequential_step_sampling: false
+    video_backend: pyav
+framework:
+  action_model:
+    action_dim: 7
+    action_hidden_dim: 2560
+    action_model_type: MLP
+    future_action_window_size: 15
+    past_action_window_size: 0
+  name: QwenOFT
+  qwenvl:
+    base_vlm: ./playground/Pretrained_models/Qwen3-VL-4B-Instruct
+output_dir: /scratch/pawsey1357/sxu1/workspace/code/starVLA/results/Checkpoints/robocasa_v02_qwenoft_cosmos50_2node_16gpu_bs8_ga1_100k
+run_id: robocasa_v02_qwenoft_cosmos50_2node_16gpu_bs8_ga1_100k
+run_root_dir: /scratch/pawsey1357/sxu1/workspace/code/starVLA/results/Checkpoints
+seed: 42
+trainer:
+  eval_interval: 1000
+  freeze_modules: ''
+  gradient_accumulation_steps: 1
+  gradient_clipping: 1.0
+  learning_rate:
+    action_model: 0.0001
+    base: 2.5e-05
+    qwen_vl_interface: 1.0e-05
+  logging_frequency: 100
+  lr_scheduler_type: cosine_with_min_lr
+  max_train_steps: 100000
+  num_warmup_steps: 100
+  optimizer:
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 1.0e-08
+  save_interval: 20000
+  scheduler_specific_kwargs:
+    min_lr: 1.0e-06
+wandb_entity: siyuhsu
+wandb_project: starVLA_robocasa_v02