results: select four-step PickupLoc history

1 files changed, 202 insertions, 0 deletions

diff --git a/results/babyai_shared/pickup_p1/h8_clean_kp.json b/results/babyai_shared/pickup_p1/h8_clean_kp.json
new file mode 100644
index 0000000..f51ec11
--- /dev/null
+++ b/results/babyai_shared/pickup_p1/h8_clean_kp.json
@@ -0,0 +1,202 @@
+{
+  "condition": "clean_kp",
+  "config": {
+    "action_dim": 7,
+    "context_gain": 1.0,
+    "hidden_layers": 4,
+    "input_dim": 7928,
+    "learning_rate": 0.03,
+    "mission_dim": 23,
+    "momentum": 0.9,
+    "reciprocal_learning_rate": 0.03,
+    "weight_decay": 0.0001,
+    "width": 256
+  },
+  "data": {
+    "color_cardinality": 6,
+    "elapsed_seconds": 81.82925486564636,
+    "env_id": "BabyAI-PickupLoc-v0",
+    "minigrid_version": "3.1.0",
+    "object_cardinality": 11,
+    "protocol": "babyai_shared_feedback_b0",
+    "rollout_episodes": 500,
+    "rollout_seed_start": 200000,
+    "state_cardinality": 3,
+    "train_episodes": 20000,
+    "train_steps": 121467,
+    "validation_episodes": 2000,
+    "validation_steps": 12347,
+    "vocabulary": [
+      "<unk>",
+      "a",
+      "ball",
+      "behind",
+      "blue",
+      "box",
+      "front",
+      "green",
+      "grey",
+      "in",
+      "key",
+      "left",
+      "of",
+      "on",
+      "pick",
+      "purple",
+      "red",
+      "right",
+      "the",
+      "up",
+      "yellow",
+      "you",
+      "your"
+    ]
+  },
+  "epoch_history": [
+    {
+      "epoch": 1,
+      "train_loss": 0.6952196103648135
+    },
+    {
+      "epoch": 2,
+      "train_loss": 0.5047668682274066
+    },
+    {
+      "epoch": 3,
+      "train_loss": 0.4749976762344963
+    },
+    {
+      "epoch": 4,
+      "train_loss": 0.4631779912898415
+    },
+    {
+      "epoch": 5,
+      "train_loss": 0.45011724729286995
+    },
+    {
+      "epoch": 6,
+      "train_loss": 0.43483660158358123
+    },
+    {
+      "epoch": 7,
+      "train_loss": 0.4248962173963848
+    },
+    {
+      "epoch": 8,
+      "train_loss": 0.4174242968935716
+    },
+    {
+      "epoch": 9,
+      "train_loss": 0.41010714556041516
+    },
+    {
+      "epoch": 10,
+      "train_loss": 0.4014566550756756
+    },
+    {
+      "epoch": 11,
+      "train_loss": 0.39507218191498206
+    },
+    {
+      "epoch": 12,
+      "train_loss": 0.3851292588836268
+    },
+    {
+      "epoch": 13,
+      "train_loss": 0.3778882456453223
+    },
+    {
+      "epoch": 14,
+      "train_loss": 0.3718921935558319
+    },
+    {
+      "epoch": 15,
+      "train_loss": 0.36200399285868595
+    },
+    {
+      "epoch": 16,
+      "train_loss": 0.3529288729868437
+    },
+    {
+      "epoch": 17,
+      "train_loss": 0.34455470483554035
+    },
+    {
+      "epoch": 18,
+      "train_loss": 0.3405523705168774
+    },
+    {
+      "epoch": 19,
+      "train_loss": 0.33339980376394174
+    },
+    {
+      "epoch": 20,
+      "train_loss": 0.32586026505420085
+    },
+    {
+      "epoch": 21,
+      "train_loss": 0.32184638095529455
+    },
+    {
+      "epoch": 22,
+      "train_loss": 0.3121714940510298
+    },
+    {
+      "epoch": 23,
+      "train_loss": 0.30236716722187246
+    },
+    {
+      "epoch": 24,
+      "train_loss": 0.2920462249768408
+    },
+    {
+      "epoch": 25,
+      "train_loss": 0.28397564828395844
+    }
+  ],
+  "epochs_completed": 25,
+  "finite": true,
+  "first_nonfinite_epoch": null,
+  "mission_lesion_rollout": {
+    "episodes": 500,
+    "mean_length": 48.154,
+    "mean_return": 0.25483437499999995,
+    "success": 0.32
+  },
+  "mission_lesion_validation": {
+    "accuracy": 0.7601846602413542,
+    "loss": 0.5676966533010533
+  },
+  "predictor": [],
+  "provenance": {
+    "cuda_device_name": "NVIDIA GeForce GTX 1080",
+    "cuda_peak_allocated_bytes": 247702528,
+    "cuda_visible_devices": "3",
+    "device": "cuda:0",
+    "git_commit": "a30c297eb8abf6f15f97dd4f35a2f530c9728100",
+    "git_dirty_tracked": false,
+    "minigrid_version": "3.1.0",
+    "torch_version": "2.3.1+cu118"
+  },
+  "rollout": {
+    "episodes": 500,
+    "mean_length": 40.478,
+    "mean_return": 0.372978125,
+    "success": 0.422
+  },
+  "stage": "babyai_pickup_p1",
+  "training": {
+    "batch_size": 256,
+    "history_steps": 8,
+    "model_seed": 4101,
+    "neutral_examples_per_epoch": 0,
+    "parameter_count_including_reciprocal_context_predictor": 2452999,
+    "shuffle_seed": 4101,
+    "total_wall_seconds_including_rollouts": 117.14025735855103,
+    "training_wall_seconds": 101.26001691818237
+  },
+  "validation": {
+    "accuracy": 0.8334818174455333,
+    "loss": 0.37354338703014944
+  }
+}