MMOneSphere: BC_PCL_PNet2_ee
This is the 'naive' way of using 'classification' PN++ to regress directly to the 3D position delta, but it seems to do a good job here despite some strange validation MSEs (sometimes they are spiking upwards). [Edit: that's because of dropout...]
Created on April 20|Last edited on May 20
Comment
(0.5744 + 0.5158 + 0.5154) / 3 = 0.5352
Train/Eval MSEs, Episode Success/Reward
Run set
3
Example GIFs
First seed after 250 epochs, looks good.
Second seed after 250 epochs, looks good.
Third seed after 250 epochs, looks good.
Variant
LGTM?
{"_hidden_keys": [],"act_type": "ee","actor_lr": 0.0001,"agent": "bc","alg_policy": "ladle_algorithmic_v02","algorithm": "BC","batch_size": 16,"bc_data_dir": "/data/dseita/softgym_mm/data_demo/MMOneSphere_v01_BClone_filtered_ladle_algorithmic_v02_nVars_200_obs_combo","bc_data_filtered": true,"data_buffer_capacity": 1000000,"encoder_type": "pointnet","env_kwargs": {"action_mode": "translation","action_repeat": 8,"camera_name": "top_down","deterministic": false,"headless": true,"horizon": 100,"num_variations": 1000,"observation_mode": "cam_rgb","render": true,"render_mode": "fluid"},"env_kwargs_camera_height": 128,"env_kwargs_camera_width": 128,"env_kwargs_deterministic": false,"env_kwargs_num_variations": 200,"env_kwargs_observation_mode": "point_cloud","env_name": "MMOneSphere","env_version": "v01","exp_name": "MMOneSphere_v01_BC_PCL_PNet2_ee_2022_04_19_23_13_42_0001","hidden_dim": 1024,"log_interval": 1,"n_epochs": 250,"n_eval_episodes": 10,"num_filters": 32,"num_layers": 4,"save_freq": 10,"save_model": true,"save_video": true,"seed": 100,"wandb_entity": "mooey5775","wandb_project": "mixed_media"}
Add a comment