benchmark_e5_xu_elim/CartPole-v1/erc_trpo/seed_0/latest checkpoint epoch 4700

benchmark_e5_xu_elim/CartPole-v1/erc_trpo/seed_0/latest/config.json

@@ -0,0 +1,49 @@
+{
+  "env_id": "CartPole-v1",
+  "env_backend": "gymnasium",
+  "results_env_id": "",
+  "variant": "erc_trpo",
+  "total_timesteps": 3000000,
+  "max_episode_steps": 500,
+  "gamma": 0.98,
+  "gae_lambda": 0.95,
+  "max_kl": 1e-05,
+  "cg_iters": 10,
+  "cg_damping": 0.1,
+  "batch_size": 500,
+  "update_batch_size": 32,
+  "entropy_coef": 1e-08,
+  "policy_hidden": [
+    128
+  ],
+  "value_hidden": [
+    128
+  ],
+  "seed": 0,
+  "num_seeds": 3,
+  "log_interval": 10,
+  "eval_interval": 10,
+  "num_eval_episodes": 10,
+  "results_dir": "results",
+  "benchmark_id": "e5_xu_elim",
+  "checkpoints_dir": "",
+  "use_buffer": false,
+  "buffer_clear_threshold": 195.0,
+  "buffer_capacity": 10000,
+  "device": "cpu",
+  "upload_to_hf": false,
+  "sync_hf_on_start": false,
+  "resume": true,
+  "save_every_epochs": 50,
+  "hf_upload_every_epochs": 0,
+  "hf_sync_interval_sec": 900,
+  "wipe_on_hyperparam_change": true,
+  "ppo_clip": 0.2,
+  "ppo_epochs": 10,
+  "ppo_lr": 0.0003,
+  "random_kl_std": 0.1,
+  "binomial_n_scale": 100.0,
+  "activation": "relu",
+  "value_lr": 0.01,
+  "value_epochs": 20
+}

benchmark_e5_xu_elim/CartPole-v1/erc_trpo/seed_0/latest/metadata.json

@@ -0,0 +1,57 @@
+{
+  "env_id": "CartPole-v1",
+  "discrete_actions": true,
+  "device": "cpu",
+  "policy_hidden": [
+    128
+  ],
+  "value_hidden": [
+    128
+  ],
+  "training": {
+    "total_timesteps": 3000000,
+    "gamma": 0.98,
+    "entropy_coef": 1e-08,
+    "seed": 0,
+    "batch_size": 500
+  },
+  "progress": {
+    "epoch": 4700,
+    "timesteps": 2350000
+  },
+  "metrics": {
+    "epoch": 4699,
+    "timesteps": 2350000,
+    "objective": 1.7485379430581816e-05,
+    "surrogate_improve": 1.7485379430581816e-05,
+    "surrogate": 1.7485379430581816e-05,
+    "kl": 5.414009137894027e-06,
+    "entropy_mean": 0.5492997765541077,
+    "kl_budget_bonus": 5.492997765541077e-09,
+    "effective_kl": 5.414009137894027e-06,
+    "line_search_backtracks": 5,
+    "update_success": 1.0,
+    "kl_relaxation": 5.492997765541077e-09,
+    "eval_return_mean": 417.2,
+    "eval_return_std": 119.44940351462621
+  },
+  "variant": "erc_trpo",
+  "uses_entropy": false,
+  "uses_buffer": false,
+  "paper": {
+    "title": "Trust region policy optimization via entropy regularization for Kullback\u2013Leibler divergence constraint",
+    "authors": "Haotian Xu, Junyu Xuan, Guangquan Zhang, Jie Lu",
+    "year": 2024,
+    "venue": "Neurocomputing 589:127716",
+    "url": "https://doi.org/10.1016/j.neucom.2024.127716"
+  },
+  "secondary_papers": [
+    {
+      "title": "Trust Region Policy Optimization",
+      "authors": "John Schulman, Sergey Levine, Philipp Moritz, Michael I. Jordan, Pieter Abbeel",
+      "year": 2015,
+      "venue": "ICML",
+      "url": ""
+    }
+  ]
+}

benchmark_e5_xu_elim/CartPole-v1/erc_trpo/seed_0/latest/policy.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f50adca4931862f851e91f02bb91bb2e4d0264aced4b3906a859e69aafff895c
+size 5971

benchmark_e5_xu_elim/CartPole-v1/erc_trpo/seed_0/latest/progress.json

@@ -0,0 +1,5 @@
+{
+  "epoch": 4700,
+  "timesteps": 2350000,
+  "history_len": 4700
+}

benchmark_e5_xu_elim/CartPole-v1/erc_trpo/seed_0/latest/value.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b27b08d6e5c2e2250460a8ab75faa19c2c8a04250331699b1fd60c139519d10e
+size 5449