Upload 291 files

First commit, upload the data sets for OpenAI Gym tasks which is processed from D4RL. Also upload all pre-trained checkpoints, including 1-ReFlow, Shortcut Policy, and DDPM trained on OpenAI Gym tasks, Franka Kitchen, and Robomimic visual manipulation tasks.

data-offline/gym_custom/ant-medium-expert-v0/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1796c4c4e0a34bbbc9710636bbbd7ec2a22d2189affed7f72ea46994d03d4e2e
+size 1970

data-offline/gym_custom/ant-medium-expert-v0/train.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d29bdcaf5f17ba311173b5206e1e23b534d125b0fb87098e722a66cf2fdb7cc3
+size 180538900

data-offline/gym_d4rl/Humanoid-medium-v3/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd1c6fdf99af3550dd259d893fdf497ca880f05b7570f08b2a65d554224ff6f3
+size 4162

data-offline/gym_d4rl/Humanoid-medium-v3/train.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e15906b8b3802639a7c6568706585b0dcf53db148e6ec2748536777ec9fb3a11
+size 393493669

data-offline/gym_d4rl/ant-medium-expert-v2/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc91a243893ee4d6d6ca562335c8c79c379b0f4f8d1ce81f572650ef84ad5b5e
+size 1970

data-offline/gym_d4rl/ant-medium-expert-v2/train.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:152c1139db6a779f92460960e28b6e319e7ee90e68ae8081ec2a2fb79bf00413
+size 171507921

data-offline/gym_d4rl/hopper-medium-v2/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d05b2943bc39772f7f770dfc0a4df2a9f205cb1e05ae74b585cd0dd382a0742e
+size 1130

data-offline/gym_d4rl/hopper-medium-v2/train.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5ae5aad98481c5fcb7a1de25536ea907a205f65f8344dc1b02e67db6957c71e
+size 37208613

data-offline/gym_d4rl/walker-medium-v2/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d103b793b591081947fca807ae13c2752bd6964621374101f5a89a19346ae16
+size 1202

data-offline/gym_d4rl/walker-medium-v2/train.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f32074e11902432f3847146890c09e8854200000103f05b9398dc38cff4f864
+size 86163981

data-offline/gym_d4rl_normalization/Humanoid-medium-v3/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd1c6fdf99af3550dd259d893fdf497ca880f05b7570f08b2a65d554224ff6f3
+size 4162

data-offline/gym_d4rl_normalization/ant-medium-expert-v2/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc91a243893ee4d6d6ca562335c8c79c379b0f4f8d1ce81f572650ef84ad5b5e
+size 1970

data-offline/gym_d4rl_normalization/hopper-medium-v2/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d05b2943bc39772f7f770dfc0a4df2a9f205cb1e05ae74b585cd0dd382a0742e
+size 1130

data-offline/gym_d4rl_normalization/walker-medium-v2/normalization.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d103b793b591081947fca807ae13c2752bd6964621374101f5a89a19346ae16
+size 1202

log/log_gym_d4rl_pretrained/Humanoid-medium-v3/1-ReFlow/state_100.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaf3a7eff02f5a0fa8a53b78ae40fc300462e18a4203e7b125712af2b98b24cb
+size 12776994

log/log_gym_d4rl_pretrained/Humanoid-medium-v3/DDPM/state_120.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:198a3343d89a78a5c0b5c62f89b49e28b64cedec90802f2b52095ae91fab5d8c
+size 12776994

log/log_gym_d4rl_pretrained/Humanoid-medium-v3/Shortcut/state_60.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4bd42a8043c8ab2b5d9d9c6309107c577a8ffeb78f014d720ab802e5485041c3
+size 10101818

log/log_gym_d4rl_pretrained/ant-medium-expert-v2/1-ReFlow/state_50.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe57f55927f3f2b10d861d9d40288f8c0396cfaafde47709019a287f779b9926
+size 10015714

log/log_gym_d4rl_pretrained/ant-medium-expert-v2/DDPM/state_20.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19a4d8374eb29758156c044b439ca92fa55af3ef652aa2c6828d3b12989611a7
+size 10015714

log/log_gym_d4rl_pretrained/ant-medium-expert-v2/Shortcut/state_50.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a590e50ded998af5ea15870ababc050408d98a0cd7374e211c5cf470b60977a6
+size 9240122

log/log_gym_d4rl_pretrained/hopper-medium-v2/1-ReFlow/state_40.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39808c2deab2e0ba275127d1b10d71098ecd747ca32f56284bd461ec6b0caccf
+size 8865250

log/log_gym_d4rl_pretrained/hopper-medium-v2/DDPM/state_120.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb512ba75abf13c61a703617b1dbe05e6ac402ad07d09b6eddb9de545f75a807
+size 8865250

log/log_gym_d4rl_pretrained/hopper-medium-v2/Shortcut/state_40.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:202a3ded34bcf5a9f592513ab8bc2765f2189986932cad6fa4c183a72ed26516
+size 8805354

log/log_gym_d4rl_pretrained/walker2d-medium-v2/1-ReFlow/state_80.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28e2c0230245ce5aed46806df614e278f39845aea68b9146169661ac020b0768
+size 9111266

log/log_gym_d4rl_pretrained/walker2d-medium-v2/DDPM/state_60.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fed53ceaf5fc10d3e51b64e4c083e7cc7577490d39d6d14f4c8f40ad3de17b8d
+size 9111266

log/log_gym_d4rl_pretrained/walker2d-medium-v2/Shortcut/state_40.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7204a4d80367467e8ce3bbdfc81922a5eee92f3dabbc2d13b939e9aaddf6539
+size 9008362

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed0/2025-04-20_20-30-16_seed0/.hydra/config.yaml

@@ -0,0 +1,129 @@
+env_name: ant-medium-expert-v0
+action_dim: 8
+horizon_steps: 4
+act_steps: 4
+obs_dim: 111
+cond_steps: 1
+_target_: agent.finetune.mine.train_ppo_flow_agent.TrainPPOFlowAgent
+name: ${env_name}_ppo_reflow_mlp_ta${horizon_steps}_td${denoising_steps}
+logdir: ${oc.env:DPPO_LOG_DIR}/gym/gym-finetune/${env_name}_ppo_reflow_mlp_ta${horizon_steps}_td${denoising_steps}_tdf${ft_denoising_steps}/${now:%Y-%m-%d}_${now:%H-%M-%S}_seed${seed}
+base_policy_path: ${oc.env:DPPO_LOG_DIR}/gym/gym-pretrain/ant-medium-expert-v0_pre_reflow_mlp_ta4_td20/2025-03-29_19-23-57_42/checkpoint/best_769.pt
+normalization_path: ${oc.env:DPPO_DATA_DIR}/gym/${env_name}/normalization.npz
+seed: 0
+device: cuda:0
+denoising_steps: 4
+ft_denoising_steps: 4
+env:
+  n_envs: 40
+  name: ${env_name}
+  max_episode_steps: 1000
+  reset_at_iteration: false
+  save_video: false
+  best_reward_threshold_for_success: 3
+  wrappers:
+    mujoco_locomotion_lowdim:
+      normalization_path: ${normalization_path}
+    multi_step:
+      n_obs_steps: ${cond_steps}
+      n_action_steps: ${act_steps}
+      max_episode_steps: ${env.max_episode_steps}
+      reset_within_step: true
+wandb:
+  entity: ${oc.env:DPPO_WANDB_ENTITY}
+  project: gym-${env_name}-finetune
+  run: ${now:%Y-%m-%d}_${now:%H-%M-%S}_${name}
+  offline_mode: false
+train:
+  n_train_itr: 1000
+  n_critic_warmup_itr: 0
+  n_steps: 500
+  gamma: 0.99
+  actor_lr: 4.5e-05
+  actor_weight_decay: 0
+  actor_lr_scheduler:
+    type: cosine
+    first_cycle_steps: 100
+    warmup_steps: 10
+    min_lr: 2.0e-05
+  critic_lr: 0.00065
+  critic_weight_decay: 1.0e-05
+  critic_lr_scheduler:
+    type: cosine
+    first_cycle_steps: 100
+    warmup_steps: 10
+    min_lr: 0.0003
+  save_model_freq: 300
+  val_freq: 10
+  render:
+    freq: 1
+    num: 0
+  reward_scale_running: true
+  reward_scale_const: 1.0
+  gae_lambda: 0.95
+  batch_size: 50000
+  update_epochs: 5
+  vf_coef: 0.5
+  ent_coef: 0.03
+  target_kl: 1.0
+  lr_schedule: fixed
+  repeat_samples: true
+  verbose: false
+  clip_intermediate_actions: true
+  account_for_initial_stochasticity: true
+model:
+  _target_: model.flow.ft_ppo.ppoflow.PPOFlow
+  device: ${device}
+  policy:
+    _target_: model.flow.mlp_flow.FlowMLP
+    horizon_steps: ${horizon_steps}
+    action_dim: ${action_dim}
+    cond_dim: ${eval:'${obs_dim} * ${cond_steps}'}
+    time_dim: 16
+    mlp_dims:
+    - 512
+    - 512
+    - 512
+    activation_type: ReLU
+    out_activation_type: Identity
+    use_layernorm: false
+    residual_style: true
+  critic:
+    _target_: model.common.critic.CriticObs
+    cond_dim: ${eval:'${obs_dim} * ${cond_steps}'}
+    mlp_dims:
+    - 256
+    - 256
+    - 256
+    activation_type: Mish
+    residual_style: true
+  actor_policy_path: ${base_policy_path}
+  act_dim: ${action_dim}
+  horizon_steps: ${horizon_steps}
+  act_min: -1
+  act_max: 1
+  obs_dim: ${obs_dim}
+  cond_steps: ${cond_steps}
+  noise_scheduler_type: learn_decay
+  inference_steps: ${denoising_steps}
+  ft_denoising_steps: ${ft_denoising_steps}
+  randn_clip_value: 3
+  min_sampling_denoising_std: 0.08
+  min_logprob_denoising_std: 0.08
+  max_logprob_denoising_std: 0.16
+  logprob_min: -1.0
+  logprob_max: 1.0
+  clip_ploss_coef: 0.01
+  clip_ploss_coef_base: 0.01
+  clip_ploss_coef_rate: 3
+  clip_vloss_coef: null
+  denoised_clip_value: 1.0
+  time_dim_explore: 0
+  learn_explore_time_embedding: false
+  use_time_independent_noise: false
+  init_time_embedding: true
+  noise_hidden_dims:
+  - 64
+  - 64
+  logprob_debug_sample: false
+  logprob_debug_recalculate: false
+  explore_net_activation_type: Tanh

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed0/2025-04-20_20-30-16_seed0/.hydra/hydra.yaml

@@ -0,0 +1,166 @@
+hydra:
+  run:
+    dir: ${logdir}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=RUN
+    task:
+    - device=cuda:0
+    - model.use_time_independent_noise=False
+    - model.noise_hidden_dims=[64,64]
+    - model.min_sampling_denoising_std=0.08
+    - model.min_logprob_denoising_std=0.08
+    - model.max_logprob_denoising_std=0.16
+    - train.ent_coef=0.03
+    - model.noise_scheduler_type=learn_decay
+    - seed=0
+  job:
+    name: run
+    chdir: null
+    override_dirname: device=cuda:0,model.max_logprob_denoising_std=0.16,model.min_logprob_denoising_std=0.08,model.min_sampling_denoising_std=0.08,model.noise_hidden_dims=[64,64],model.noise_scheduler_type=learn_decay,model.use_time_independent_noise=False,seed=0,train.ent_coef=0.03
+    id: ???
+    num: ???
+    config_name: ft_ppo_reflow_mlp_learn
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/zhangtonghe/dppo
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/zhangtonghe/dppo/cfg
+      schema: file
+      provider: main
+    - path: /home/zhangtonghe/dppo/cfg/gym/finetune/ant-medium-expert-v0
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/zhangtonghe/dppo/log/gym/gym-finetune/ant-medium-expert-v0_ppo_reflow_mlp_ta4_td4_tdf4/2025-04-20_20-30-16_seed0
+    choices:
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed0/2025-04-20_20-30-16_seed0/.hydra/overrides.yaml

@@ -0,0 +1,9 @@
+- device=cuda:0
+- model.use_time_independent_noise=False
+- model.noise_hidden_dims=[64,64]
+- model.min_sampling_denoising_std=0.08
+- model.min_logprob_denoising_std=0.08
+- model.max_logprob_denoising_std=0.16
+- train.ent_coef=0.03
+- model.noise_scheduler_type=learn_decay
+- seed=0

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed0/2025-04-20_20-30-16_seed0/architecture.log

@@ -0,0 +1,76 @@
+self.model=PPOFlow(
+  (actor_old): FlowMLP(
+    (time_embedding): Sequential(
+      (0): SinusoidalPosEmb()
+      (1): Linear(in_features=16, out_features=32, bias=True)
+      (2): Mish()
+      (3): Linear(in_features=32, out_features=16, bias=True)
+    )
+    (mlp_mean): ResidualMLP(
+      (layers): ModuleList(
+        (0): Linear(in_features=159, out_features=512, bias=True)
+        (1): TwoLayerPreActivationResNetLinear(
+          (l1): Linear(in_features=512, out_features=512, bias=True)
+          (l2): Linear(in_features=512, out_features=512, bias=True)
+          (act): ReLU()
+        )
+        (2): Linear(in_features=512, out_features=32, bias=True)
+        (3): Identity()
+      )
+    )
+  )
+  (actor_ft): NoisyFlowMLP(
+    (policy): FlowMLP(
+      (time_embedding): Sequential(
+        (0): SinusoidalPosEmb()
+        (1): Linear(in_features=16, out_features=32, bias=True)
+        (2): Mish()
+        (3): Linear(in_features=32, out_features=16, bias=True)
+      )
+      (mlp_mean): ResidualMLP(
+        (layers): ModuleList(
+          (0): Linear(in_features=159, out_features=512, bias=True)
+          (1): TwoLayerPreActivationResNetLinear(
+            (l1): Linear(in_features=512, out_features=512, bias=True)
+            (l2): Linear(in_features=512, out_features=512, bias=True)
+            (act): ReLU()
+          )
+          (2): Linear(in_features=512, out_features=32, bias=True)
+          (3): Identity()
+        )
+      )
+    )
+    (explore_noise_net): ExploreNoiseNet(
+      (mlp_logvar): MLP(
+        (moduleList): ModuleList(
+          (0): Sequential(
+            (linear_1): Linear(in_features=127, out_features=64, bias=True)
+            (act_1): Tanh()
+          )
+          (1): Sequential(
+            (linear_1): Linear(in_features=64, out_features=64, bias=True)
+            (act_1): Tanh()
+          )
+          (2): Sequential(
+            (linear_1): Linear(in_features=64, out_features=32, bias=True)
+            (act_1): Identity()
+          )
+        )
+      )
+    )
+  )
+  (critic): CriticObs(
+    (Q1): ResidualMLP(
+      (layers): ModuleList(
+        (0): Linear(in_features=111, out_features=256, bias=True)
+        (1): TwoLayerPreActivationResNetLinear(
+          (l1): Linear(in_features=256, out_features=256, bias=True)
+          (l2): Linear(in_features=256, out_features=256, bias=True)
+          (act): Mish()
+        )
+        (2): Linear(in_features=256, out_features=1, bias=True)
+        (3): Identity()
+      )
+    )
+  )
+)

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed0/2025-04-20_20-30-16_seed0/checkpoint/best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c9e47be0089acf12e3c654b3e08cddd1acc3de311b8a0816724341c562a8392
+size 12129240

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed0/2025-04-20_20-30-16_seed0/checkpoint/state_999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12b211f06a3356b8b43262635bd2ff2944554c8e504b5865996663cf13e74077
+size 12135418

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed0/2025-04-20_20-30-16_seed0/result.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17f93a4b41b51f4d13867f3a73a9f3ac3470a10fb8ec7d3fd2838dcba665d0fa
+size 845615

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed3407/2025-04-20_20-30-25_seed3407/.hydra/config.yaml

@@ -0,0 +1,129 @@
+env_name: ant-medium-expert-v0
+action_dim: 8
+horizon_steps: 4
+act_steps: 4
+obs_dim: 111
+cond_steps: 1
+_target_: agent.finetune.mine.train_ppo_flow_agent.TrainPPOFlowAgent
+name: ${env_name}_ppo_reflow_mlp_ta${horizon_steps}_td${denoising_steps}
+logdir: ${oc.env:DPPO_LOG_DIR}/gym/gym-finetune/${env_name}_ppo_reflow_mlp_ta${horizon_steps}_td${denoising_steps}_tdf${ft_denoising_steps}/${now:%Y-%m-%d}_${now:%H-%M-%S}_seed${seed}
+base_policy_path: ${oc.env:DPPO_LOG_DIR}/gym/gym-pretrain/ant-medium-expert-v0_pre_reflow_mlp_ta4_td20/2025-03-29_19-23-57_42/checkpoint/best_769.pt
+normalization_path: ${oc.env:DPPO_DATA_DIR}/gym/${env_name}/normalization.npz
+seed: 42
+device: cuda:0
+denoising_steps: 4
+ft_denoising_steps: 4
+env:
+  n_envs: 40
+  name: ${env_name}
+  max_episode_steps: 1000
+  reset_at_iteration: false
+  save_video: false
+  best_reward_threshold_for_success: 3
+  wrappers:
+    mujoco_locomotion_lowdim:
+      normalization_path: ${normalization_path}
+    multi_step:
+      n_obs_steps: ${cond_steps}
+      n_action_steps: ${act_steps}
+      max_episode_steps: ${env.max_episode_steps}
+      reset_within_step: true
+wandb:
+  entity: ${oc.env:DPPO_WANDB_ENTITY}
+  project: gym-${env_name}-finetune
+  run: ${now:%Y-%m-%d}_${now:%H-%M-%S}_${name}
+  offline_mode: false
+train:
+  n_train_itr: 1000
+  n_critic_warmup_itr: 0
+  n_steps: 500
+  gamma: 0.99
+  actor_lr: 4.5e-05
+  actor_weight_decay: 0
+  actor_lr_scheduler:
+    type: cosine
+    first_cycle_steps: 100
+    warmup_steps: 10
+    min_lr: 2.0e-05
+  critic_lr: 0.00065
+  critic_weight_decay: 1.0e-05
+  critic_lr_scheduler:
+    type: cosine
+    first_cycle_steps: 100
+    warmup_steps: 10
+    min_lr: 0.0003
+  save_model_freq: 300
+  val_freq: 10
+  render:
+    freq: 1
+    num: 0
+  reward_scale_running: true
+  reward_scale_const: 1.0
+  gae_lambda: 0.95
+  batch_size: 50000
+  update_epochs: 5
+  vf_coef: 0.5
+  ent_coef: 0.03
+  target_kl: 1.0
+  lr_schedule: fixed
+  repeat_samples: true
+  verbose: false
+  clip_intermediate_actions: true
+  account_for_initial_stochasticity: true
+model:
+  _target_: model.flow.ft_ppo.ppoflow.PPOFlow
+  device: ${device}
+  policy:
+    _target_: model.flow.mlp_flow.FlowMLP
+    horizon_steps: ${horizon_steps}
+    action_dim: ${action_dim}
+    cond_dim: ${eval:'${obs_dim} * ${cond_steps}'}
+    time_dim: 16
+    mlp_dims:
+    - 512
+    - 512
+    - 512
+    activation_type: ReLU
+    out_activation_type: Identity
+    use_layernorm: false
+    residual_style: true
+  critic:
+    _target_: model.common.critic.CriticObs
+    cond_dim: ${eval:'${obs_dim} * ${cond_steps}'}
+    mlp_dims:
+    - 256
+    - 256
+    - 256
+    activation_type: Mish
+    residual_style: true
+  actor_policy_path: ${base_policy_path}
+  act_dim: ${action_dim}
+  horizon_steps: ${horizon_steps}
+  act_min: -1
+  act_max: 1
+  obs_dim: ${obs_dim}
+  cond_steps: ${cond_steps}
+  noise_scheduler_type: learn_decay
+  inference_steps: ${denoising_steps}
+  ft_denoising_steps: ${ft_denoising_steps}
+  randn_clip_value: 3
+  min_sampling_denoising_std: 0.08
+  min_logprob_denoising_std: 0.08
+  max_logprob_denoising_std: 0.16
+  logprob_min: -1.0
+  logprob_max: 1.0
+  clip_ploss_coef: 0.01
+  clip_ploss_coef_base: 0.01
+  clip_ploss_coef_rate: 3
+  clip_vloss_coef: null
+  denoised_clip_value: 1.0
+  time_dim_explore: 0
+  learn_explore_time_embedding: false
+  use_time_independent_noise: false
+  init_time_embedding: true
+  noise_hidden_dims:
+  - 64
+  - 64
+  logprob_debug_sample: false
+  logprob_debug_recalculate: false
+  explore_net_activation_type: Tanh

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed3407/2025-04-20_20-30-25_seed3407/.hydra/hydra.yaml

@@ -0,0 +1,166 @@
+hydra:
+  run:
+    dir: ${logdir}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=RUN
+    task:
+    - device=cuda:1
+    - model.use_time_independent_noise=False
+    - model.noise_hidden_dims=[64,64]
+    - model.min_sampling_denoising_std=0.08
+    - model.min_logprob_denoising_std=0.08
+    - model.max_logprob_denoising_std=0.16
+    - train.ent_coef=0.03
+    - model.noise_scheduler_type=learn_decay
+    - seed=3407
+  job:
+    name: run
+    chdir: null
+    override_dirname: device=cuda:1,model.max_logprob_denoising_std=0.16,model.min_logprob_denoising_std=0.08,model.min_sampling_denoising_std=0.08,model.noise_hidden_dims=[64,64],model.noise_scheduler_type=learn_decay,model.use_time_independent_noise=False,seed=3407,train.ent_coef=0.03
+    id: ???
+    num: ???
+    config_name: ft_ppo_reflow_mlp_learn
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/zhangtonghe/dppo
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/zhangtonghe/dppo/cfg
+      schema: file
+      provider: main
+    - path: /home/zhangtonghe/dppo/cfg/gym/finetune/ant-medium-expert-v0
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/zhangtonghe/dppo/log/gym/gym-finetune/ant-medium-expert-v0_ppo_reflow_mlp_ta4_td4_tdf4/2025-04-20_20-30-25_seed3407
+    choices:
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed3407/2025-04-20_20-30-25_seed3407/.hydra/overrides.yaml

@@ -0,0 +1,9 @@
+- device=cuda:1
+- model.use_time_independent_noise=False
+- model.noise_hidden_dims=[64,64]
+- model.min_sampling_denoising_std=0.08
+- model.min_logprob_denoising_std=0.08
+- model.max_logprob_denoising_std=0.16
+- train.ent_coef=0.03
+- model.noise_scheduler_type=learn_decay
+- seed=3407

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed3407/2025-04-20_20-30-25_seed3407/architecture.log

@@ -0,0 +1,76 @@
+self.model=PPOFlow(
+  (actor_old): FlowMLP(
+    (time_embedding): Sequential(
+      (0): SinusoidalPosEmb()
+      (1): Linear(in_features=16, out_features=32, bias=True)
+      (2): Mish()
+      (3): Linear(in_features=32, out_features=16, bias=True)
+    )
+    (mlp_mean): ResidualMLP(
+      (layers): ModuleList(
+        (0): Linear(in_features=159, out_features=512, bias=True)
+        (1): TwoLayerPreActivationResNetLinear(
+          (l1): Linear(in_features=512, out_features=512, bias=True)
+          (l2): Linear(in_features=512, out_features=512, bias=True)
+          (act): ReLU()
+        )
+        (2): Linear(in_features=512, out_features=32, bias=True)
+        (3): Identity()
+      )
+    )
+  )
+  (actor_ft): NoisyFlowMLP(
+    (policy): FlowMLP(
+      (time_embedding): Sequential(
+        (0): SinusoidalPosEmb()
+        (1): Linear(in_features=16, out_features=32, bias=True)
+        (2): Mish()
+        (3): Linear(in_features=32, out_features=16, bias=True)
+      )
+      (mlp_mean): ResidualMLP(
+        (layers): ModuleList(
+          (0): Linear(in_features=159, out_features=512, bias=True)
+          (1): TwoLayerPreActivationResNetLinear(
+            (l1): Linear(in_features=512, out_features=512, bias=True)
+            (l2): Linear(in_features=512, out_features=512, bias=True)
+            (act): ReLU()
+          )
+          (2): Linear(in_features=512, out_features=32, bias=True)
+          (3): Identity()
+        )
+      )
+    )
+    (explore_noise_net): ExploreNoiseNet(
+      (mlp_logvar): MLP(
+        (moduleList): ModuleList(
+          (0): Sequential(
+            (linear_1): Linear(in_features=127, out_features=64, bias=True)
+            (act_1): Tanh()
+          )
+          (1): Sequential(
+            (linear_1): Linear(in_features=64, out_features=64, bias=True)
+            (act_1): Tanh()
+          )
+          (2): Sequential(
+            (linear_1): Linear(in_features=64, out_features=32, bias=True)
+            (act_1): Identity()
+          )
+        )
+      )
+    )
+  )
+  (critic): CriticObs(
+    (Q1): ResidualMLP(
+      (layers): ModuleList(
+        (0): Linear(in_features=111, out_features=256, bias=True)
+        (1): TwoLayerPreActivationResNetLinear(
+          (l1): Linear(in_features=256, out_features=256, bias=True)
+          (l2): Linear(in_features=256, out_features=256, bias=True)
+          (act): Mish()
+        )
+        (2): Linear(in_features=256, out_features=1, bias=True)
+        (3): Identity()
+      )
+    )
+  )
+)

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed3407/2025-04-20_20-30-25_seed3407/checkpoint/best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2ac28e2c3de5ac5a6c88ae5d8424488a168669b265e27da2fc849bf90783821
+size 12129240

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed3407/2025-04-20_20-30-25_seed3407/checkpoint/state_999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a9490fb1e3d3c926896f2f497bba017655fdf8a1249e10cea8685e26c8d3e3c
+size 12135418

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed3407/2025-04-20_20-30-25_seed3407/result.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bab79232bd796b4ada6e144181b9241593e0193ba814a030ad25f032f445936
+size 845615

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed42/2025-04-20_10-14-30_seed42/.hydra/config.yaml

@@ -0,0 +1,129 @@
+env_name: ant-medium-expert-v0
+action_dim: 8
+horizon_steps: 4
+act_steps: 4
+obs_dim: 111
+cond_steps: 1
+_target_: agent.finetune.mine.train_ppo_flow_agent.TrainPPOFlowAgent
+name: ${env_name}_ppo_reflow_mlp_ta${horizon_steps}_td${denoising_steps}
+logdir: ${oc.env:DPPO_LOG_DIR}/gym/gym-finetune/${env_name}_ppo_reflow_mlp_ta${horizon_steps}_td${denoising_steps}_tdf${ft_denoising_steps}/${now:%Y-%m-%d}_${now:%H-%M-%S}_seed${seed}
+base_policy_path: ${oc.env:DPPO_LOG_DIR}/gym/gym-pretrain/ant-medium-expert-v0_pre_reflow_mlp_ta4_td20/2025-03-29_19-23-57_42/checkpoint/best_769.pt
+normalization_path: ${oc.env:DPPO_DATA_DIR}/gym/${env_name}/normalization.npz
+seed: 42
+device: cuda:0
+denoising_steps: 4
+ft_denoising_steps: 4
+env:
+  n_envs: 40
+  name: ${env_name}
+  max_episode_steps: 1000
+  reset_at_iteration: false
+  save_video: false
+  best_reward_threshold_for_success: 3
+  wrappers:
+    mujoco_locomotion_lowdim:
+      normalization_path: ${normalization_path}
+    multi_step:
+      n_obs_steps: ${cond_steps}
+      n_action_steps: ${act_steps}
+      max_episode_steps: ${env.max_episode_steps}
+      reset_within_step: true
+wandb:
+  entity: ${oc.env:DPPO_WANDB_ENTITY}
+  project: gym-${env_name}-finetune
+  run: ${now:%Y-%m-%d}_${now:%H-%M-%S}_${name}
+  offline_mode: false
+train:
+  n_train_itr: 1000
+  n_critic_warmup_itr: 0
+  n_steps: 500
+  gamma: 0.99
+  actor_lr: 4.5e-05
+  actor_weight_decay: 0
+  actor_lr_scheduler:
+    type: cosine
+    first_cycle_steps: 100
+    warmup_steps: 10
+    min_lr: 2.0e-05
+  critic_lr: 0.00065
+  critic_weight_decay: 1.0e-05
+  critic_lr_scheduler:
+    type: cosine
+    first_cycle_steps: 100
+    warmup_steps: 10
+    min_lr: 0.0003
+  save_model_freq: 300
+  val_freq: 10
+  render:
+    freq: 1
+    num: 0
+  reward_scale_running: true
+  reward_scale_const: 1.0
+  gae_lambda: 0.95
+  batch_size: 50000
+  update_epochs: 5
+  vf_coef: 0.5
+  ent_coef: 0.03
+  target_kl: 1.0
+  lr_schedule: fixed
+  repeat_samples: true
+  verbose: false
+  clip_intermediate_actions: true
+  account_for_initial_stochasticity: true
+model:
+  _target_: model.flow.ft_ppo.ppoflow.PPOFlow
+  device: ${device}
+  policy:
+    _target_: model.flow.mlp_flow.FlowMLP
+    horizon_steps: ${horizon_steps}
+    action_dim: ${action_dim}
+    cond_dim: ${eval:'${obs_dim} * ${cond_steps}'}
+    time_dim: 16
+    mlp_dims:
+    - 512
+    - 512
+    - 512
+    activation_type: ReLU
+    out_activation_type: Identity
+    use_layernorm: false
+    residual_style: true
+  critic:
+    _target_: model.common.critic.CriticObs
+    cond_dim: ${eval:'${obs_dim} * ${cond_steps}'}
+    mlp_dims:
+    - 256
+    - 256
+    - 256
+    activation_type: Mish
+    residual_style: true
+  actor_policy_path: ${base_policy_path}
+  act_dim: ${action_dim}
+  horizon_steps: ${horizon_steps}
+  act_min: -1
+  act_max: 1
+  obs_dim: ${obs_dim}
+  cond_steps: ${cond_steps}
+  noise_scheduler_type: learn_decay
+  inference_steps: ${denoising_steps}
+  ft_denoising_steps: ${ft_denoising_steps}
+  randn_clip_value: 3
+  min_sampling_denoising_std: 0.08
+  min_logprob_denoising_std: 0.08
+  max_logprob_denoising_std: 0.16
+  logprob_min: -1.0
+  logprob_max: 1.0
+  clip_ploss_coef: 0.01
+  clip_ploss_coef_base: 0.01
+  clip_ploss_coef_rate: 3
+  clip_vloss_coef: null
+  denoised_clip_value: 1.0
+  time_dim_explore: 0
+  learn_explore_time_embedding: false
+  use_time_independent_noise: false
+  init_time_embedding: true
+  noise_hidden_dims:
+  - 64
+  - 64
+  logprob_debug_sample: false
+  logprob_debug_recalculate: false
+  explore_net_activation_type: Tanh

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed42/2025-04-20_10-14-30_seed42/.hydra/hydra.yaml

@@ -0,0 +1,165 @@
+hydra:
+  run:
+    dir: ${logdir}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=RUN
+    task:
+    - device=cuda:3
+    - model.use_time_independent_noise=False
+    - model.noise_hidden_dims=[64,64]
+    - model.min_sampling_denoising_std=0.08
+    - model.min_logprob_denoising_std=0.08
+    - model.max_logprob_denoising_std=0.16
+    - train.ent_coef=0.03
+    - model.noise_scheduler_type=learn_decay
+  job:
+    name: run
+    chdir: null
+    override_dirname: device=cuda:3,model.max_logprob_denoising_std=0.16,model.min_logprob_denoising_std=0.08,model.min_sampling_denoising_std=0.08,model.noise_hidden_dims=[64,64],model.noise_scheduler_type=learn_decay,model.use_time_independent_noise=False,train.ent_coef=0.03
+    id: ???
+    num: ???
+    config_name: ft_ppo_reflow_mlp_learn
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/zhangtonghe/dppo
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/zhangtonghe/dppo/cfg
+      schema: file
+      provider: main
+    - path: /home/zhangtonghe/dppo/cfg/gym/finetune/ant-medium-expert-v0
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/zhangtonghe/dppo/log/gym/gym-finetune/ant-medium-expert-v0_ppo_reflow_mlp_ta4_td4_tdf4/2025-04-20_10-14-30_seed42
+    choices:
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed42/2025-04-20_10-14-30_seed42/.hydra/overrides.yaml

@@ -0,0 +1,8 @@
+- device=cuda:3
+- model.use_time_independent_noise=False
+- model.noise_hidden_dims=[64,64]
+- model.min_sampling_denoising_std=0.08
+- model.min_logprob_denoising_std=0.08
+- model.max_logprob_denoising_std=0.16
+- train.ent_coef=0.03
+- model.noise_scheduler_type=learn_decay

log/log_gym_finetuned/gym/ant-v0/ReFlow/seed42/2025-04-20_10-14-30_seed42/architecture.log

@@ -0,0 +1,76 @@
+self.model=PPOFlow(
+  (actor_old): FlowMLP(
+    (time_embedding): Sequential(
+      (0): SinusoidalPosEmb()
+      (1): Linear(in_features=16, out_features=32, bias=True)
+      (2): Mish()
+      (3): Linear(in_features=32, out_features=16, bias=True)
+    )
+    (mlp_mean): ResidualMLP(
+      (layers): ModuleList(
+        (0): Linear(in_features=159, out_features=512, bias=True)
+        (1): TwoLayerPreActivationResNetLinear(
+          (l1): Linear(in_features=512, out_features=512, bias=True)
+          (l2): Linear(in_features=512, out_features=512, bias=True)
+          (act): ReLU()
+        )
+        (2): Linear(in_features=512, out_features=32, bias=True)
+        (3): Identity()
+      )
+    )
+  )
+  (actor_ft): NoisyFlowMLP(
+    (policy): FlowMLP(
+      (time_embedding): Sequential(
+        (0): SinusoidalPosEmb()
+        (1): Linear(in_features=16, out_features=32, bias=True)
+        (2): Mish()
+        (3): Linear(in_features=32, out_features=16, bias=True)
+      )
+      (mlp_mean): ResidualMLP(
+        (layers): ModuleList(
+          (0): Linear(in_features=159, out_features=512, bias=True)
+          (1): TwoLayerPreActivationResNetLinear(
+            (l1): Linear(in_features=512, out_features=512, bias=True)
+            (l2): Linear(in_features=512, out_features=512, bias=True)
+            (act): ReLU()
+          )
+          (2): Linear(in_features=512, out_features=32, bias=True)
+          (3): Identity()
+        )
+      )
+    )
+    (explore_noise_net): ExploreNoiseNet(
+      (mlp_logvar): MLP(
+        (moduleList): ModuleList(
+          (0): Sequential(
+            (linear_1): Linear(in_features=127, out_features=64, bias=True)
+            (act_1): Tanh()
+          )
+          (1): Sequential(
+            (linear_1): Linear(in_features=64, out_features=64, bias=True)
+            (act_1): Tanh()
+          )
+          (2): Sequential(
+            (linear_1): Linear(in_features=64, out_features=32, bias=True)
+            (act_1): Identity()
+          )
+        )
+      )
+    )
+  )
+  (critic): CriticObs(
+    (Q1): ResidualMLP(
+      (layers): ModuleList(
+        (0): Linear(in_features=111, out_features=256, bias=True)
+        (1): TwoLayerPreActivationResNetLinear(
+          (l1): Linear(in_features=256, out_features=256, bias=True)
+          (l2): Linear(in_features=256, out_features=256, bias=True)
+          (act): Mish()
+        )
+        (2): Linear(in_features=256, out_features=1, bias=True)
+        (3): Identity()
+      )
+    )
+  )
+)