diff --git a/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/seed42_my_atomic26_nenv10/RowWiseAdaptiveProbMix.yaml b/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/seed42_my_atomic26_nenv10/RowWiseAdaptiveProbMix.yaml new file mode 100644 index 0000000000000000000000000000000000000000..687b3966bc9b9244a0d5ce37deb016b947437f5b --- /dev/null +++ b/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/seed42_my_atomic26_nenv10/RowWiseAdaptiveProbMix.yaml @@ -0,0 +1,55 @@ +experiment_name: DTW from 60k - RowWiseAdaptiveProbMix +policy_type: groot_DTW +base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-60000 +output_root: ${HOME}/groot_robocasa/robocasa_v2/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 + seed: 42 +model: + dtw_enabled: true + dtw_relation_mode: flatten + dtw_loss_type: kl + dtw_action_temp: 0.1 + dtw_student_temp: 0.1 + dtw_student_reconstruction_head_enabled: true + dtw_student_reconstruction_head_dim: 512 + dtw_student_reconstruction_head_hidden_dim: 512 + dtw_student_reconstruction_head_pooling: attention + dtw_mix_mode: rowwise_adaptive_prob_mix + dtw_pos_weight: 1.0 + dtw_rot_weight: 1.0 + dtw_gripper_weight: 1.0 + dtw_base_weight: 1.0 + dtw_control_weight: 1.0 + dtw_normalize_components: true + dtw_adaptive_mix_lambda: 1.0 + dtw_adaptive_floor: 0.01 + dtw_timing_log_interval: 1000 + dtw_exclude_diagonal: true +phases: +- name: phase2_kl_dtw_aphead_rowwise_adaptive_only + max_steps: 30000 + save_steps: 0 + trainable: + preset: processing_line_only + tune_llm: false + tune_visual: false + tune_projector: false + tune_diffusion_model: false + losses: + dtw_fm_loss_weight: 0.0 + dtw_kl_loss_weight: 1.0 +- name: phase3_fm_kl_dtw_aphead_rowwise_adaptive_w0p5_f0p01 + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + dtw_fm_loss_weight: 1.0 + dtw_kl_loss_weight: 0.5 diff --git a/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/seed42_my_atomic26_nenv10/eval.log b/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/seed42_my_atomic26_nenv10/eval.log new file mode 100644 index 0000000000000000000000000000000000000000..e683eccfd731d1fcfbf8afdb1a9545d6d840f979 --- /dev/null +++ b/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/seed42_my_atomic26_nenv10/eval.log @@ -0,0 +1,5535 @@ +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/seonho/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +WARNING: mimicgen environments not imported since mimicgen is not installed! +/home/seonho/miniconda3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. +Model not found or avail in the huggingface hub. Loading from local path: /home/seonho/groot_robocasa/robocasa_v2/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/default/phase3 +Loading pretrained dual brain from /home/seonho/groot_robocasa/robocasa_v2/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/default/phase3 +Tune backbone vision tower: True +Tune backbone LLM: False +Tune action head projector: True +Tune action head DiT: True +Model not found or avail in the huggingface hub. Loading from local path: /home/seonho/groot_robocasa/robocasa_v2/DTW/DTW_v1/from_60k/RowWiseAdaptiveProbMix/default/phase3 +Available modality configs: +Tune backbone llm: False +Tune backbone visual: True +Total number of DiT parameters: 550386688 +Total number of SelfAttentionTransformer parameters: 201433088 +Tune action head projector: True +Tune action head diffusion model: True + Loading checkpoint shards: 0%| | 0/2 [00:00