Merge pull request #4642 from buckman-google/master

Addition of STEVE

research/steve/config/envs/walker2d.json

+{
+  "updates": {
+    "env": {
+      "name": "Walker2d-v1",
+      "obs_dims": [17],
+      "action_dim": 6,
+      "reward_scale":1.0,
+      "discount":0.99,
+      "max_frames": 1000
+    }
+  }
+}
\ No newline at end of file

research/steve/config/experimental_setups/speedrun.json

+{
+  "updates": {
+    "policy_config": {
+      "frames_per_update": false
+    },
+    "*model_config":{
+      "frames_per_update": false,
+      "pretrain_n": false
+    }
+  }
+}
\ No newline at end of file

research/steve/config/experiments/ablations/baselines/ensemble_mve_tdk0.json

+{"inherits": ["config/algos/mve_tdk.json", "config/core/bayesian", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/ensemble_mve_tdk1.json

+{"inherits": ["config/algos/mve_tdk.json", "config/core/bayesian", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/ensemble_mve_tdk2.json

+{"inherits": ["config/algos/mve_tdk.json", "config/core/bayesian", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/mve_25tdlambda0.json

+{
+  "inherits": ["config/algos/mve_tdlambda.json", "config/envs/humanoid.json"],
+  "updates":{
+    "policy_config": {
+      "value_expansion": {
+        "lambda_return": 0.25
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/baselines/mve_25tdlambda1.json

+{
+  "inherits": ["config/algos/mve_tdlambda.json", "config/envs/humanoid.json"],
+  "updates":{
+    "policy_config": {
+      "value_expansion": {
+        "lambda_return": 0.25
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/baselines/mve_25tdlambda2.json

+{
+  "inherits": ["config/algos/mve_tdlambda.json", "config/envs/humanoid.json"],
+  "updates":{
+    "policy_config": {
+      "value_expansion": {
+        "lambda_return": 0.25
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/baselines/mve_75tdlambda0.json

+{
+  "inherits": ["config/algos/mve_tdlambda.json", "config/envs/humanoid.json"],
+  "updates":{
+    "policy_config": {
+      "value_expansion": {
+        "lambda_return": 0.75
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/baselines/mve_75tdlambda1.json

+{
+  "inherits": ["config/algos/mve_tdlambda.json", "config/envs/humanoid.json"],
+  "updates":{
+    "policy_config": {
+      "value_expansion": {
+        "lambda_return": 0.75
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/baselines/mve_75tdlambda2.json

+{
+  "inherits": ["config/algos/mve_tdlambda.json", "config/envs/humanoid.json"],
+  "updates":{
+    "policy_config": {
+      "value_expansion": {
+        "lambda_return": 0.75
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/baselines/mve_meank0.json

+{"inherits": ["config/algos/mve_mean.json", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/mve_meank1.json

+{"inherits": ["config/algos/mve_mean.json", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/mve_meank2.json

+{"inherits": ["config/algos/mve_mean.json", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/steve_cov0.json

+{"inherits": ["config/algos/steve_cov.json", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/steve_cov1.json

+{"inherits": ["config/algos/steve_cov.json", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/baselines/steve_cov2.json

+{"inherits": ["config/algos/steve_cov.json", "config/envs/humanoid.json"]}

research/steve/config/experiments/ablations/horizons/steve_1h0.json

+{
+  "inherits": ["config/algos/steve.json", "config/envs/humanoid.json"],
+  "updates": {
+    "policy_config": {
+      "value_expansion": {
+        "rollout_len": 1
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/horizons/steve_1h1.json

+{
+  "inherits": ["config/algos/steve.json", "config/envs/humanoid.json"],
+  "updates": {
+    "policy_config": {
+      "value_expansion": {
+        "rollout_len": 1
+      }
+    }
+  }
+}

research/steve/config/experiments/ablations/horizons/steve_1h2.json

+{
+  "inherits": ["config/algos/steve.json", "config/envs/humanoid.json"],
+  "updates": {
+    "policy_config": {
+      "value_expansion": {
+        "rollout_len": 1
+      }
+    }
+  }
+}