update

Real-world/checkpoints/VLA-JEPA-Real-World.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1ab32b71d2c5f2f755bdb42d9815dece05446d288cbf38bd946c01981949779
+size 6163571823

Real-world/config.json

@@ -0,0 +1,114 @@
+{
+  "run_id": "fr3_realworld",
+  "run_root_dir": "checkpoints",
+  "seed": 42,
+  "trackers": [
+    "json"
+  ],
+  "is_debug": false,
+  "framework": {
+    "name": "VLA_JEPA",
+    "qwenvl": {
+      "base_vlm": "/home/dataset-local/models/Qwen3-VL-2B-Instruct",
+      "attn_implementation": "flash_attention_2",
+      "vl_hidden_dim": 2048
+    },
+    "action_model": {
+      "action_model_type": "DiT-B",
+      "action_hidden_dim": 1024,
+      "hidden_size": 1024,
+      "add_pos_embed": true,
+      "max_seq_len": 1024,
+      "action_dim": 7,
+      "state_dim": 8,
+      "future_action_window_size": 6,
+      "action_horizon": 7,
+      "past_action_window_size": 0,
+      "repeated_diffusion_steps": 8,
+      "noise_beta_alpha": 1.5,
+      "noise_beta_beta": 1.0,
+      "noise_s": 0.999,
+      "num_timestep_buckets": 1000,
+      "num_inference_timesteps": 4,
+      "num_target_vision_tokens": 32,
+      "diffusion_model_cfg": {
+        "cross_attention_dim": 2048,
+        "dropout": 0.2,
+        "final_dropout": true,
+        "interleave_self_attention": true,
+        "norm_type": "ada_norm",
+        "num_layers": 16,
+        "output_dim": 1024,
+        "positional_embeddings": null
+      }
+    },
+    "vj2_model": {
+      "base_encoder": "/home/dataset-local/models/vjepa2-vitl-fpc64-256",
+      "depth": 12,
+      "num_heads": 8,
+      "special_action_token": "<|action_{}|>",
+      "num_action_tokens_per_timestep": 8,
+      "embodied_action_token": "<|embodied_action|>",
+      "num_embodied_action_tokens_per_instruction": 32,
+      "num_frames": 8
+    },
+    "reduce_in_full_precision": true
+  },
+  "datasets": {
+    "vla_data": {
+      "dataset_py": "lerobot_datasets",
+      "data_root_dir": "/home/dataset-local/datasets/LeRobot/lerobot_simple_pp_starvla",
+      "data_mix": "fr3_realworld",
+      "action_type": "delta_qpos",
+      "CoT_prompt": "Your task is {instruction}. Infer the temporal dynamics from frames {actions} and produce the corresponding policy actions {e_actions}.",
+      "resolution_size": 224,
+      "per_device_batch_size": 8,
+      "video_resolution_size": 256,
+      "load_all_data_for_training": true,
+      "with_state": true
+    }
+  },
+  "trainer": {
+    "epochs": 100,
+    "max_train_steps": 20000,
+    "num_warmup_steps": 5000,
+    "save_interval": 5000,
+    "eval_interval": 100,
+    "learning_rate": {
+      "base": 3e-05,
+      "qwen_vl_interface": 1e-05,
+      "action_model": 0.0001
+    },
+    "lr_scheduler_type": "cosine_with_min_lr",
+    "scheduler_specific_kwargs": {
+      "min_lr": 1e-06
+    },
+    "freeze_modules": "",
+    "loss_scale": {
+      "vla": 1.0,
+      "vlm": 0.1
+    },
+    "max_grad_norm": 1.0,
+    "warmup_ratio": 0.1,
+    "weight_decay": 0.0,
+    "logging_frequency": 10,
+    "gradient_clipping": 1.0,
+    "gradient_accumulation_steps": 1,
+    "pretrained_checkpoint": "/home/dataset-local/VLA_JEPA/checkpoints/pretrain/VLA-JEPA-pretrain.pt",
+    "optimizer": {
+      "name": "AdamW",
+      "betas": [
+        0.9,
+        0.95
+      ],
+      "eps": 1e-08,
+      "weight_decay": 1e-08
+    },
+    "is_resume": false,
+    "resume_epoch": null,
+    "resume_step": null,
+    "enable_gradient_checkpointing": true,
+    "enable_mixed_precision_training": true
+  },
+  "output_dir": "checkpoints/fr3_realworld"
+}

Real-world/config.yaml

@@ -0,0 +1,99 @@
+run_id: fr3_realworld
+run_root_dir: checkpoints
+seed: 42
+trackers:
+- json
+is_debug: false
+framework:
+  name: VLA_JEPA
+  qwenvl:
+    base_vlm: /home/dataset-local/models/Qwen3-VL-2B-Instruct
+    attn_implementation: flash_attention_2
+    vl_hidden_dim: 2048
+  action_model:
+    action_model_type: DiT-B
+    action_hidden_dim: 1024
+    hidden_size: 1024
+    add_pos_embed: true
+    max_seq_len: 1024
+    action_dim: 7
+    state_dim: 8
+    future_action_window_size: 6
+    action_horizon: 7
+    past_action_window_size: 0
+    repeated_diffusion_steps: 8
+    noise_beta_alpha: 1.5
+    noise_beta_beta: 1.0
+    noise_s: 0.999
+    num_timestep_buckets: 1000
+    num_inference_timesteps: 4
+    num_target_vision_tokens: 32
+    diffusion_model_cfg:
+      cross_attention_dim: 2048
+      dropout: 0.2
+      final_dropout: true
+      interleave_self_attention: true
+      norm_type: ada_norm
+      num_layers: 16
+      output_dim: 1024
+      positional_embeddings: null
+  vj2_model:
+    base_encoder: /home/dataset-local/models/vjepa2-vitl-fpc64-256
+    depth: 12
+    num_heads: 8
+    special_action_token: <|action_{}|>
+    num_action_tokens_per_timestep: 8
+    embodied_action_token: <|embodied_action|>
+    num_embodied_action_tokens_per_instruction: 32
+    num_frames: 8
+  reduce_in_full_precision: true
+datasets:
+  vla_data:
+    dataset_py: lerobot_datasets
+    data_root_dir: /home/dataset-local/datasets/LeRobot/lerobot_simple_pp_starvla
+    data_mix: fr3_realworld
+    action_type: delta_qpos
+    CoT_prompt: Your task is {instruction}. Infer the temporal dynamics from frames
+      {actions} and produce the corresponding policy actions {e_actions}.
+    resolution_size: 224
+    per_device_batch_size: 8
+    video_resolution_size: 256
+    load_all_data_for_training: true
+    with_state: true
+trainer:
+  epochs: 100
+  max_train_steps: 20000
+  num_warmup_steps: 5000
+  save_interval: 5000
+  eval_interval: 100
+  learning_rate:
+    base: 3.0e-05
+    qwen_vl_interface: 1.0e-05
+    action_model: 0.0001
+  lr_scheduler_type: cosine_with_min_lr
+  scheduler_specific_kwargs:
+    min_lr: 1.0e-06
+  freeze_modules: ''
+  loss_scale:
+    vla: 1.0
+    vlm: 0.1
+  max_grad_norm: 1.0
+  warmup_ratio: 0.1
+  weight_decay: 0.0
+  logging_frequency: 10
+  gradient_clipping: 1.0
+  gradient_accumulation_steps: 1
+  pretrained_checkpoint: /home/dataset-local/VLA_JEPA/checkpoints/pretrain/VLA-JEPA-pretrain.pt
+  optimizer:
+    name: AdamW
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 1.0e-08
+  is_resume: false
+  resume_epoch: null
+  resume_step: null
+  enable_gradient_checkpointing: true
+  enable_mixed_precision_training: true
+output_dir: checkpoints/fr3_realworld

Real-world/dataset_statistics.json

@@ -0,0 +1,133 @@
+{
+  "new_embodiment": {
+    "action": {
+      "mean": [
+        -1.6495379895786755e-05,
+        -1.1559173799469136e-05,
+        6.800065875722794e-06,
+        -2.9312453989405185e-05,
+        2.715121809160337e-05,
+        -4.9356358431396075e-06,
+        0.8368043303489685
+      ],
+      "std": [
+        0.0033433015923947096,
+        0.0033241035416722298,
+        0.006203544791787863,
+        0.0064756181091070175,
+        0.006977501790970564,
+        0.008858172222971916,
+        0.308319091796875
+      ],
+      "max": [
+        0.029447495937347412,
+        0.04054729640483856,
+        0.05029946565628052,
+        0.04862421378493309,
+        0.08689296990633011,
+        0.06699639558792114,
+        1.0
+      ],
+      "min": [
+        -0.022726356983184814,
+        -0.027304204180836678,
+        -0.04183477163314819,
+        -0.06220978870987892,
+        -0.07182798534631729,
+        -0.09515094757080078,
+        0.0
+      ],
+      "q01": [
+        -0.009029481410980224,
+        -0.010530177503824234,
+        -0.014956550300121307,
+        -0.02058939129114151,
+        -0.020688764695078136,
+        -0.03448918495327234,
+        0.0
+      ],
+      "q99": [
+        0.010799104869365696,
+        0.01133852298371494,
+        0.018483443856239404,
+        0.01734598506242038,
+        0.019764822572469743,
+        0.022766803707927472,
+        1.0
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        false
+      ]
+    },
+    "state": {
+      "mean": [
+        0.564656674861908,
+        -0.002990487962961197,
+        0.3436369001865387,
+        0.8995662331581116,
+        -0.32835933566093445,
+        0.03922935202717781,
+        -0.01520092599093914,
+        0.8368043303489685
+      ],
+      "std": [
+        0.087661974132061,
+        0.09329120814800262,
+        0.16795401275157928,
+        0.18832752108573914,
+        0.1940889209508896,
+        0.07308010756969452,
+        0.051968540996313095,
+        0.308319091796875
+      ],
+      "max": [
+        0.8323888182640076,
+        0.2805258333683014,
+        0.7719749808311462,
+        0.9999507069587708,
+        0.8582825660705566,
+        0.3295547068119049,
+        0.17514149844646454,
+        1.0
+      ],
+      "min": [
+        0.2107870727777481,
+        -0.3139313757419586,
+        0.06993856281042099,
+        -0.7052522301673889,
+        -0.7020109295845032,
+        -0.24225156009197235,
+        -0.22992828488349915,
+        0.0
+      ],
+      "q01": [
+        0.29528580099344254,
+        -0.27157875895500183,
+        0.08648627504706383,
+        -0.5524575877189636,
+        -0.6196129459142685,
+        -0.12603880420327188,
+        -0.1619066223502159,
+        0.0
+      ],
+      "q99": [
+        0.7763650172948837,
+        0.22109754353761682,
+        0.5853771680593494,
+        0.9986419814825058,
+        0.7496437698602678,
+        0.24052399903535857,
+        0.11271571815013894,
+        1.0
+      ]
+    },
+    "num_transitions": 17919,
+    "num_trajectories": 100
+  }
+}

Real-world/summary.jsonl

@@ -0,0 +1,6 @@
+{"steps": 5000}
+{"steps": 10000}
+{"steps": 15000}
+{"steps": 20000}
+{"steps": 25000}
+{"steps": 30000}

SimplerEnv/checkpoints/VLA-JEPA-SimplerEnv.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f797a93590847960cbd14feb56c29e5fbc39220c162ed29451c385caa9dab6e
+size 6163573444

SimplerEnv/config.json

@@ -0,0 +1,115 @@
+{
+  "run_id": "SimplerEnv",
+  "run_root_dir": "checkpoints",
+  "seed": 42,
+  "trackers": [
+    "json"
+  ],
+  "is_debug": false,
+  "framework": {
+    "name": "VLA_JEPA",
+    "qwenvl": {
+      "base_vlm": "/home/dataset-local/models/Qwen3-VL-2B-Instruct",
+      "attn_implementation": "flash_attention_2",
+      "vl_hidden_dim": 2048
+    },
+    "action_model": {
+      "action_model_type": "DiT-B",
+      "action_hidden_dim": 1024,
+      "hidden_size": 1024,
+      "add_pos_embed": true,
+      "max_seq_len": 1024,
+      "action_dim": 7,
+      "state_dim": 8,
+      "future_action_window_size": 6,
+      "action_horizon": 7,
+      "past_action_window_size": 0,
+      "repeated_diffusion_steps": 8,
+      "noise_beta_alpha": 1.5,
+      "noise_beta_beta": 1.0,
+      "noise_s": 0.999,
+      "num_timestep_buckets": 1000,
+      "num_inference_timesteps": 4,
+      "num_target_vision_tokens": 32,
+      "diffusion_model_cfg": {
+        "cross_attention_dim": 2048,
+        "dropout": 0.2,
+        "final_dropout": true,
+        "interleave_self_attention": true,
+        "norm_type": "ada_norm",
+        "num_layers": 16,
+        "output_dim": 1024,
+        "positional_embeddings": null
+      }
+    },
+    "vj2_model": {
+      "base_encoder": "/home/dataset-local/models/vjepa2-vitl-fpc64-256",
+      "depth": 12,
+      "num_heads": 8,
+      "special_action_token": "<|action_{}|>",
+      "num_action_tokens_per_timestep": 8,
+      "embodied_action_token": "<|embodied_action|>",
+      "num_embodied_action_tokens_per_instruction": 32,
+      "num_frames": 8
+    },
+    "reduce_in_full_precision": true
+  },
+  "datasets": {
+    "vla_data": {
+      "dataset_py": "lerobot_datasets",
+      "data_root_dir": "/home/dataset-local/datasets/LeRobot/OXE_LEROBOT_DATASET",
+      "data_mix": "bridge_rt_1",
+      "action_type": "delta_ee",
+      "CoT_prompt": "Your task is {instruction}. Infer the temporal dynamics from frames {actions} and produce the corresponding policy actions {e_actions}.",
+      "resolution_size": 224,
+      "video_resolution_size": 256,
+      "per_device_batch_size": 32,
+      "load_all_data_for_training": true,
+      "action_horizon": 7,
+      "with_state": false
+    }
+  },
+  "trainer": {
+    "epochs": 100,
+    "max_train_steps": 30000,
+    "num_warmup_steps": 5000,
+    "save_interval": 10000,
+    "eval_interval": 100,
+    "learning_rate": {
+      "base": 3e-05,
+      "qwen_vl_interface": 1e-05,
+      "action_model": 0.0001,
+      "vj_predictor": 0.0005
+    },
+    "lr_scheduler_type": "cosine_with_min_lr",
+    "scheduler_specific_kwargs": {
+      "min_lr": 1e-05
+    },
+    "freeze_modules": "",
+    "loss_scale": {
+      "vla": 1.0,
+      "vlm": 0.1
+    },
+    "max_grad_norm": 1.0,
+    "warmup_ratio": 0.1,
+    "weight_decay": 0.0,
+    "logging_frequency": 10,
+    "gradient_clipping": 1.0,
+    "gradient_accumulation_steps": 1,
+    "optimizer": {
+      "name": "AdamW",
+      "betas": [
+        0.9,
+        0.95
+      ],
+      "eps": 1e-08,
+      "weight_decay": 1e-08
+    },
+    "is_resume": false,
+    "resume_epoch": null,
+    "resume_step": null,
+    "enable_gradient_checkpointing": true,
+    "enable_mixed_precision_training": true
+  },
+  "output_dir": "checkpoints/SimplerEnv"
+}

SimplerEnv/config.yaml

@@ -0,0 +1,100 @@
+run_id: SimplerEnv
+run_root_dir: checkpoints
+seed: 42
+trackers:
+- json
+is_debug: false
+framework:
+  name: VLA_JEPA
+  qwenvl:
+    base_vlm: /home/dataset-local/models/Qwen3-VL-2B-Instruct
+    attn_implementation: flash_attention_2
+    vl_hidden_dim: 2048
+  action_model:
+    action_model_type: DiT-B
+    action_hidden_dim: 1024
+    hidden_size: 1024
+    add_pos_embed: true
+    max_seq_len: 1024
+    action_dim: 7
+    state_dim: 8
+    future_action_window_size: 6
+    action_horizon: 7
+    past_action_window_size: 0
+    repeated_diffusion_steps: 8
+    noise_beta_alpha: 1.5
+    noise_beta_beta: 1.0
+    noise_s: 0.999
+    num_timestep_buckets: 1000
+    num_inference_timesteps: 4
+    num_target_vision_tokens: 32
+    diffusion_model_cfg:
+      cross_attention_dim: 2048
+      dropout: 0.2
+      final_dropout: true
+      interleave_self_attention: true
+      norm_type: ada_norm
+      num_layers: 16
+      output_dim: 1024
+      positional_embeddings: null
+  vj2_model:
+    base_encoder: /home/dataset-local/models/vjepa2-vitl-fpc64-256
+    depth: 12
+    num_heads: 8
+    special_action_token: <|action_{}|>
+    num_action_tokens_per_timestep: 8
+    embodied_action_token: <|embodied_action|>
+    num_embodied_action_tokens_per_instruction: 32
+    num_frames: 8
+  reduce_in_full_precision: true
+datasets:
+  vla_data:
+    dataset_py: lerobot_datasets
+    data_root_dir: /home/dataset-local/datasets/LeRobot/OXE_LEROBOT_DATASET
+    data_mix: bridge_rt_1
+    action_type: delta_ee
+    CoT_prompt: Your task is {instruction}. Infer the temporal dynamics from frames
+      {actions} and produce the corresponding policy actions {e_actions}.
+    resolution_size: 224
+    video_resolution_size: 256
+    per_device_batch_size: 32
+    load_all_data_for_training: true
+    action_horizon: 7
+    with_state: false
+trainer:
+  epochs: 100
+  max_train_steps: 30000
+  num_warmup_steps: 5000
+  save_interval: 10000
+  eval_interval: 100
+  learning_rate:
+    base: 3.0e-05
+    qwen_vl_interface: 1.0e-05
+    action_model: 0.0001
+    vj_predictor: 0.0005
+  lr_scheduler_type: cosine_with_min_lr
+  scheduler_specific_kwargs:
+    min_lr: 1.0e-05
+  freeze_modules: ''
+  loss_scale:
+    vla: 1.0
+    vlm: 0.1
+  max_grad_norm: 1.0
+  warmup_ratio: 0.1
+  weight_decay: 0.0
+  logging_frequency: 10
+  gradient_clipping: 1.0
+  gradient_accumulation_steps: 1
+  optimizer:
+    name: AdamW
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 1.0e-08
+  is_resume: false
+  resume_epoch: null
+  resume_step: null
+  enable_gradient_checkpointing: true
+  enable_mixed_precision_training: true
+output_dir: checkpoints/SimplerEnv

SimplerEnv/dataset_statistics.json

@@ -0,0 +1,264 @@
+{
+  "oxe_bridge": {
+    "action": {
+      "mean": [
+        0.00011365935642970726,
+        6.556110747624189e-05,
+        -6.320965621853247e-05,
+        -7.205353176686913e-05,
+        -0.00019515302847139537,
+        0.0001203166029881686,
+        0.28829458355903625
+      ],
+      "std": [
+        0.006909770731857718,
+        0.009684093232852218,
+        0.00896290498528129,
+        0.020121052930683073,
+        0.021582655517295487,
+        0.054723342223346974,
+        0.4543627821514982
+      ],
+      "max": [
+        0.41691166162490845,
+        0.25864794850349426,
+        0.21218234300613403,
+        3.122201919555664,
+        1.8618112802505493,
+        6.272472858428955,
+        1.0
+      ],
+      "min": [
+        -0.4007510244846344,
+        -0.13874775171279907,
+        -0.22553899884223938,
+        -3.2010786533355713,
+        -1.8618112802505493,
+        -6.279075622558594,
+        0.0
+      ],
+      "q01": [
+        -0.02875255048274994,
+        -0.041702136397361755,
+        -0.02609672024846077,
+        -0.08052875101566315,
+        -0.09249906986951828,
+        -0.20738555490970612,
+        0.0
+      ],
+      "q99": [
+        0.028306663036346436,
+        0.040898531675338745,
+        0.0401805154979229,
+        0.08173403143882751,
+        0.07760760188102722,
+        0.2038465440273285,
+        1.0
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        false
+      ]
+    },
+    "state": {
+      "mean": [
+        0.15471743047237396,
+        0.015362550504505634,
+        0.03222028166055679,
+        0.0032453201711177826,
+        -0.038600146770477295,
+        0.05382946878671646,
+        0.0,
+        0.35406652092933655
+      ],
+      "std": [
+        0.1605359274864927,
+        0.06677912092232112,
+        0.048657150951527554,
+        0.09275336958940797,
+        0.12566984746914736,
+        0.41222738578801793,
+        0.0,
+        0.4330223535305803
+      ],
+      "max": [
+        0.5862360596656799,
+        0.4034728705883026,
+        0.3568263053894043,
+        1.3517684936523438,
+        1.570796251296997,
+        3.141204357147217,
+        0.0,
+        1.1121242046356201
+      ],
+      "min": [
+        -0.04167502000927925,
+        -0.3563207685947418,
+        -0.15537554025650024,
+        -3.141592502593994,
+        -1.4992541074752808,
+        -3.14153790473938,
+        0.0,
+        0.04637829214334488
+      ],
+      "q01": [
+        0.17102622985839844,
+        -0.1698118895292282,
+        -0.05563216283917427,
+        -0.36493173241615295,
+        -0.541871190071106,
+        -1.3542754650115967,
+        0.0,
+        0.052190229296684265
+      ],
+      "q99": [
+        0.45322078466415405,
+        0.2354845106601715,
+        0.19489620625972748,
+        0.3780156075954437,
+        0.27568644285202026,
+        1.8500566482543945,
+        0.0,
+        1.0105689764022827
+      ]
+    },
+    "num_transitions": 1863900,
+    "num_trajectories": 53192
+  },
+  "oxe_rt1": {
+    "action": {
+      "mean": [
+        0.003493750700727105,
+        0.003132961690425873,
+        -0.0063125672750175,
+        0.02166595682501793,
+        -0.0028780836146324873,
+        0.0004565489653032273,
+        0.26771023869514465
+      ],
+      "std": [
+        0.049065014465362655,
+        0.04229853739828572,
+        0.05237628880142378,
+        0.1124860236500875,
+        0.09312952783816872,
+        0.10319098309601091,
+        0.4418448662622395
+      ],
+      "max": [
+        2.9984593391418457,
+        22.09052848815918,
+        2.7507524490356445,
+        1.570636510848999,
+        1.5321086645126343,
+        1.5691522359848022,
+        1.0
+      ],
+      "min": [
+        -2.0204520225524902,
+        -5.497899532318115,
+        -2.031663417816162,
+        -1.569917917251587,
+        -1.569892168045044,
+        -1.570419430732727,
+        0.0
+      ],
+      "q01": [
+        -0.224535271525383,
+        -0.1482001394033432,
+        -0.23158970475196838,
+        -0.35179948806762695,
+        -0.4193011224269867,
+        -0.43643462657928467,
+        0.0
+      ],
+      "q99": [
+        0.17824687063694,
+        0.1493837833404541,
+        0.21842354536056519,
+        0.5892665982246399,
+        0.352726548910141,
+        0.4479667842388153,
+        1.0
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        false
+      ]
+    },
+    "state": {
+      "mean": [
+        0.2799473702907562,
+        -0.04167069122195244,
+        0.38854750990867615,
+        -0.12402277439832687,
+        0.24756911396980286,
+        0.046330634504556656,
+        0.10487449914216995,
+        0.21306729316711426
+      ],
+      "std": [
+        0.29342642876909925,
+        0.09174024655686211,
+        0.42569508885539115,
+        0.38314586427420927,
+        0.44433568806919804,
+        0.1263927443679382,
+        0.22122596673781084,
+        0.38616252611341306
+      ],
+      "max": [
+        1.0534898042678833,
+        0.48018959164619446,
+        1.6896663904190063,
+        0.9999993443489075,
+        0.9999874830245972,
+        0.9554369449615479,
+        0.9914546012878418,
+        1.0
+      ],
+      "min": [
+        -0.4436439275741577,
+        -0.9970501065254211,
+        -0.006579156965017319,
+        -0.8643477559089661,
+        -0.7079970240592957,
+        -0.7688722014427185,
+        -0.4999994933605194,
+        0.0
+      ],
+      "q01": [
+        0.3248138129711151,
+        -0.2833428978919983,
+        0.14107070863246918,
+        -0.6864742040634155,
+        -0.6808923482894897,
+        -0.3604559600353241,
+        -0.45438095927238464,
+        0.0
+      ],
+      "q99": [
+        0.8750156164169312,
+        0.21247053146362305,
+        1.0727112293243408,
+        0.9377871155738831,
+        0.9563050866127014,
+        0.4599004089832306,
+        0.7216041088104248,
+        1.0
+      ]
+    },
+    "num_transitions": 3449894,
+    "num_trajectories": 87212
+  }
+}

SimplerEnv/summary.jsonl

@@ -0,0 +1,3 @@
+{"steps": 10000}
+{"steps": 20000}
+{"steps": 30000}