model: graph_conv-transe_h128_l3_edge_prediction | (graph_conv-transe_h128_l3) | WandB: i3kmoqkx

config.json

@@ -2,14 +2,14 @@
   "name": null,
   "seed": 42,
   "deterministic": true,
-  "output_dir": "relation_prediction_transe_128e_transfer_v1",
+  "output_dir": "relation_prediction_transe_128e_transfer_v2",
   "model": {
     "encoder": "graph_conv",
     "hidden_channels": 128,
     "num_layers": 3,
     "dropout": 0.2,
     "head": "transe",
-    "init_head_as_identity": true,
+    "init_head_as_identity": false,
     "gat_heads": 4,
     "gat_concat": true,
     "graph_conv_aggregator": "mean",
@@ -28,9 +28,9 @@
     "edge_encoder_init_bias": 0.0,
     "use_pretrained_model": true,
     "pretrained_model_source": "huggingface",
-    "pretrained_model_path": "seanhacks/edge_prediction_dotprod_128e",
-    "pretrained_model_revision": null,
-    "pretrained_model_load_head": false,
+    "pretrained_model_path": "seanhacks/relation_prediction_transe_128e",
+    "pretrained_model_revision": "20251228",
+    "pretrained_model_load_head": true,
     "pretrained_model_freeze_encoder_weights": false,
     "pretrained_model_freeze_head_weights": false
   },
@@ -63,7 +63,7 @@
     "lr": 0.0005,
     "weight_decay": 0.0001,
     "optimizer": "adamw",
-    "scheduler": "onecycle",
+    "scheduler": "plateau",
     "gradient_clip_val": null,
     "epochs": 500,
     "batches_per_epoch": 20,
@@ -71,7 +71,7 @@
     "devices": 1,
     "precision": 32,
     "early_stopping": true,
-    "early_stopping_patience": 50,
+    "early_stopping_patience": 100,
     "early_stopping_metric": "val_relation_weighted_auc",
     "save_checkpoints": true,
     "checkpoint_subdir": "checkpoints",