results: reject diagonal SDIL on PickupLoc

1 files changed, 295 insertions, 0 deletions

diff --git a/results/babyai_shared/pickup_p2/seed4102_sdil.json b/results/babyai_shared/pickup_p2/seed4102_sdil.json
new file mode 100644
index 0000000..d06ebe4
--- /dev/null
+++ b/results/babyai_shared/pickup_p2/seed4102_sdil.json
@@ -0,0 +1,295 @@
+{
+  "condition": "sdil",
+  "config": {
+    "action_dim": 7,
+    "context_gain": 1.0,
+    "hidden_layers": 4,
+    "input_dim": 3964,
+    "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": 1.4037305317426982
+    },
+    {
+      "epoch": 2,
+      "train_loss": 1.5199073093815854
+    },
+    {
+      "epoch": 3,
+      "train_loss": 1.490153865814209
+    },
+    {
+      "epoch": 4,
+      "train_loss": 1.3832739104722676
+    },
+    {
+      "epoch": 5,
+      "train_loss": 1.343873203428168
+    },
+    {
+      "epoch": 6,
+      "train_loss": 1.5590096739718788
+    },
+    {
+      "epoch": 7,
+      "train_loss": 1.6177833466780813
+    },
+    {
+      "epoch": 8,
+      "train_loss": 1.3586675656469245
+    },
+    {
+      "epoch": 9,
+      "train_loss": 1.4770930380570262
+    },
+    {
+      "epoch": 10,
+      "train_loss": 1.4020695854488172
+    },
+    {
+      "epoch": 11,
+      "train_loss": 1.660883920067235
+    },
+    {
+      "epoch": 12,
+      "train_loss": 1.4256328241448653
+    },
+    {
+      "epoch": 13,
+      "train_loss": 1.3678666669444033
+    },
+    {
+      "epoch": 14,
+      "train_loss": 1.5727304014406707
+    },
+    {
+      "epoch": 15,
+      "train_loss": 1.4565586441441587
+    },
+    {
+      "epoch": 16,
+      "train_loss": 1.461268330373262
+    },
+    {
+      "epoch": 17,
+      "train_loss": 1.39766593155108
+    },
+    {
+      "epoch": 18,
+      "train_loss": 1.4142666897020841
+    },
+    {
+      "epoch": 19,
+      "train_loss": 1.476966627020585
+    },
+    {
+      "epoch": 20,
+      "train_loss": 1.4425141741100111
+    },
+    {
+      "epoch": 21,
+      "train_loss": 1.457439170134695
+    },
+    {
+      "epoch": 22,
+      "train_loss": 1.3950570189325433
+    },
+    {
+      "epoch": 23,
+      "train_loss": 1.387892014101932
+    },
+    {
+      "epoch": 24,
+      "train_loss": 1.5007062272021645
+    },
+    {
+      "epoch": 25,
+      "train_loss": 1.4848456322519403
+    },
+    {
+      "epoch": 26,
+      "train_loss": 1.3709611882661519
+    },
+    {
+      "epoch": 27,
+      "train_loss": 1.4066063529566715
+    },
+    {
+      "epoch": 28,
+      "train_loss": 1.5965287012802927
+    },
+    {
+      "epoch": 29,
+      "train_loss": 1.4587098232068514
+    },
+    {
+      "epoch": 30,
+      "train_loss": 1.5121034484160574
+    },
+    {
+      "epoch": 31,
+      "train_loss": 1.551018009436758
+    },
+    {
+      "epoch": 32,
+      "train_loss": 1.4567223797346416
+    },
+    {
+      "epoch": 33,
+      "train_loss": 1.3996574670390078
+    },
+    {
+      "epoch": 34,
+      "train_loss": 1.3792390657726088
+    },
+    {
+      "epoch": 35,
+      "train_loss": 1.5840236104162115
+    },
+    {
+      "epoch": 36,
+      "train_loss": 1.4191108206698768
+    },
+    {
+      "epoch": 37,
+      "train_loss": 1.5500210498508653
+    },
+    {
+      "epoch": 38,
+      "train_loss": 1.8080971825750252
+    },
+    {
+      "epoch": 39,
+      "train_loss": 1.5819644702108282
+    },
+    {
+      "epoch": 40,
+      "train_loss": 1.5324563553458765
+    }
+  ],
+  "epochs_completed": 40,
+  "finite": true,
+  "first_nonfinite_epoch": null,
+  "mission_lesion_rollout": {
+    "episodes": 500,
+    "mean_length": 64.0,
+    "mean_return": 0.0,
+    "success": 0.0
+  },
+  "mission_lesion_validation": {
+    "accuracy": 0.49890661699198185,
+    "loss": 1.2736983379198104
+  },
+  "predictor": [
+    {
+      "action_observations": 0,
+      "context_rms": 0.21022486686706543,
+      "mean_per_cell_r2": 0.14354562759399414,
+      "neutral_observations": 4096,
+      "residual_context_rms_ratio": 0.6005861759185791,
+      "teaching_observations": 0
+    },
+    {
+      "action_observations": 0,
+      "context_rms": 0.20877140760421753,
+      "mean_per_cell_r2": 0.3952368497848511,
+      "neutral_observations": 4096,
+      "residual_context_rms_ratio": 0.5114675164222717,
+      "teaching_observations": 0
+    },
+    {
+      "action_observations": 0,
+      "context_rms": 0.20698049664497375,
+      "mean_per_cell_r2": 0.3069768249988556,
+      "neutral_observations": 4096,
+      "residual_context_rms_ratio": 0.5448850393295288,
+      "teaching_observations": 0
+    },
+    {
+      "action_observations": 0,
+      "context_rms": 0.20679785311222076,
+      "mean_per_cell_r2": 0.3696891665458679,
+      "neutral_observations": 4096,
+      "residual_context_rms_ratio": 0.5242287516593933,
+      "teaching_observations": 0
+    }
+  ],
+  "provenance": {
+    "cuda_device_name": "NVIDIA GeForce GTX 1080",
+    "cuda_peak_allocated_bytes": 435783168,
+    "cuda_visible_devices": "7",
+    "device": "cuda:0",
+    "git_commit": "7b7d7a08a5e78c3379397f01f0c9af97d5d84fbd",
+    "git_dirty_tracked": false,
+    "minigrid_version": "3.1.0",
+    "torch_version": "2.3.1+cu118"
+  },
+  "rollout": {
+    "episodes": 500,
+    "mean_length": 64.0,
+    "mean_return": 0.0,
+    "success": 0.0
+  },
+  "stage": "babyai_pickup_p2",
+  "training": {
+    "batch_size": 256,
+    "history_steps": 4,
+    "model_seed": 4102,
+    "neutral_examples_per_epoch": 4096,
+    "parameter_count_including_reciprocal_context_predictor": 1438215,
+    "shuffle_seed": 4102,
+    "total_wall_seconds_including_rollouts": 344.80195808410645,
+    "training_wall_seconds": 318.03023290634155
+  },
+  "validation": {
+    "accuracy": 0.49890661699198185,
+    "loss": 1.2736801557756392
+  }
+}